Computers and Electrical Engineering

Size: px
Start display at page:

Download "Computers and Electrical Engineering"

Transcription

1 Computers and Electrical Engineering 39 (2013) Contents lists available at SciVerse ScienceDirect Computers and Electrical Engineering journal homepage: Automated two-dimensional K-means clustering algorithm for unsupervised image segmentation q Intan Aidha Yusoff, Nor Ashidi Mat Isa, Khairunnisa Hasikin Imaging and Intelligent System Research Team (ISRT), School of Electrical & Electronic Engineering, Engineering Campus, Universiti Sains Malaysia, Nibong Tebal, Pulau Pinang, Malaysia article info abstract Article history: Available online 11 December 2012 This paper introduces the Automated Two-Dimensional K-Means (A2DKM) algorithm, a novel unsupervised clustering technique. The proposed technique differs from the conventional clustering techniques because it eliminates the need for users to determine the number of clusters. In addition, A2DKM incorporates local and spatial information of the data into the clustering analysis. A2DKM is qualitatively and quantitatively compared with the conventional clustering algorithms, namely, the K-Means (KM), Fuzzy C-Means (FCM), Moving K-Means (MKM), and Adaptive Fuzzy K-Means (AFKM) algorithms. The A2DKM outperforms these algorithms by producing more homogeneous segmentation results. Ó 2012 Elsevier Ltd. All rights reserved. 1. Introduction The vast array of computer vision products in the market has caused a rise in the number of computer vision and digital image processing applications. Segmentation is one of the stages in image processing, wherein an image is simplified by segmenting it into several meaningful regions. These segmented regions are essential for assisting tasks on image processing and analysis. Image segmentation in computer vision has applications in areas such as agricultural and food inspection [1 3], document processing [5], photography [5 7], and display or acquisition devices [8]. The most common segmentation approaches have a wide variety of applications, which include template matching [9], thresholding [10], region growing [11], and clustering [12,13]. The aforementioned approaches, especially the clustering algorithms, show the significance of image segmentation in the development of consumer electronic products. Thus, this paper aims to further contribute towards the application of clustering techniques in consumer electronics by introducing a new clustering algorithm. Clustering is the process wherein image pixels are grouped into clusters, such that cluster members have certain notable similarities. Several conventional clustering algorithms are available for implementation. Among which, the K-Means (KM), Fuzzy C-Means (FCM), and Moving K-Means (MKM) algorithms are the most widely used, having been continuously perfected over time. The KM clustering algorithm clusters data into a fixed number of clusters using the Euclidean distance based on the pixel s intensity value [14], whereas the FCM clustering algorithm allows data to become a member of more than one cluster but with different membership degrees [15]. The MKM clustering algorithm improves KM clustering by combining the fitness concept with the basic concept of the original algorithm [16]. One of the more recently developed clustering techniques is the Adaptive Fuzzy K-Means (AFKM) clustering algorithm [17]. The AFKM algorithm incorporates the conventional approaches of the KM, FCM, and MKM algorithms into an improved q Reviews processed and recommended for publication to Editor-in-Chief by Deputy Editor Dr. Ferat Sahin. Corresponding author. Tel.: ; fax: addresses: intan_aidha@yahoo.com (I.A. Yusoff), ashidi@eng.usm.my (N.A. Mat Isa), eishahasikin@gmail.com (K. Hasikin) /$ - see front matter Ó 2012 Elsevier Ltd. All rights reserved.

2 908 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) clustering technique. The KM and MKM concepts are used to assign the data to its nearest center, whereas the FCM concept is included in this approach to allow the data to belong to two or more clusters [17]. The conventional clustering algorithms (i.e., the KM, FCM, and MKM clustering algorithms) and the AFKM clustering algorithm have the following common limitations. The number of clusters for segmentation needs to be estimated and defined by the user at the initial stage of the segmentation process. Clustering is performed based on a single local parameter, namely, the intensity value of a pixel, thereby discarding the correlative effects of spatial parameters on a pixel, which leads to information loss during segmentation. Therefore, new clustering algorithms that do not depend on a user-defined number of clusters are needed to overcome these limitations. Furthermore, the new clustering algorithms should incorporate the spatial information of data into its decision making process. The prior determination of the number of clusters requires an extra step in preparing the images for clustering. As for example, given that different users may define a different number of clusters for the same image. No specific method can guarantee the optimum number of clusters for that particular image. Thus, this manual procedure may be time-consuming and subjective. Numerous studies have been conducted on developing unsupervised segmentation algorithms to help ease the implementation of clustering applications over the years [18 20]. In addition to the development and enhancement of clustering algorithm performance, several researchers have integrated spatial characteristics during image clustering. This approach is intended to reduce information loss and noisy pixel interference during segmentation [21 23]. The intensity or local average intensity of a pixel (as the spatial parameter) commonly used by the researchers has been adapted into the FCM clustering algorithm. Therefore, the Automated Two-Dimensional K-Means (A2DKM) clustering algorithm is developed in this study to overcome the two aforementioned limitations. The main motivation of the new clustering technique is to build an unsupervised clustering algorithm which automatically determines the optimum number of clusters for a noiseless image by using the local and spatial characteristics of the input image. This technique will eliminate the need to declare the number of clusters prior to clustering, especially for images where the optimum number of clusters is unknown. Conventionally, the concept of the intensity median value in a specific N N window of an image is adopted in noise filtering techniques to reduce unwanted noise. The proposed A2DKM algorithm in this study is originally designed to segment noiseless images, which is contrary to the removal of unwanted noise in the image that is performed by noise filtering techniques. The concept of segmentation is specifically employed to remove insignificant details which occur in the regions to be segmented. The insignificant details refer to any edges, small regions, among others, which are ideally not segmented as independent regions. Such details include the wave patterns on sea or lake water surfaces, brick edges on building walls, and the different colors of tree leaves. These details exist in the form of a sudden change in the gradients of an intensity level. Thus, the concept of the intensity median value in a specific N N window (i.e., not solely dependent on the individual intensity level) that is employed in this study can possibly reduce and/or remove the effects of such details during segmentation. The removal of these insignificant details believed to provide better segmentation results by producing more homogeneous characteristics in their respective regions. The rest of this paper is organized as follows: The proposed clustering algorithm is explained in Section 2. Section 3 illustrates the proposed algorithm as it is implemented on a standard image. Section 4 explains the methods used for data analysis. The results obtained from the proposed A2DKM algorithm are analyzed in Section 5 and its performance is evaluated and compared with that of several selected conventional and relatively new clustering algorithms. Finally, Section 6 concludes the work presented in this paper. 2. Proposed approach In segmenting an 8-bit gray scale digital image with 256 gray levels within the interval [0, 255] using the proposed A2DKM algorithm, let p(x,y) be the pixel at location (x,y) with an intensity p (where x =1,2,3,..., R and y =1,2,3,..., S, with R and S as the number of columns and rows of the image, respectively). The initial values for all clusters are predetermined, and the number of clusters is initially set to two. All the data are then assigned to the nearest center based on the Euclidean distance. Generally, the conventional KM clustering algorithm will minimize the following objective function for partitioning a dataset fv t g N t¼1 into the kth center (c k) with a predefined number of clusters (n c ) [24]: J ¼ Xnc X N k¼1 t¼1 kv t c k k 2 ð1þ where kk stands for a distance measurement that is normally taken to be the Euclidean norm. The proposed A2DKM clustering algorithm differs from the KM clustering algorithm by including the contribution of the local spatial information of data. Thus, v t and c k in (1) are represented by (2) and (3), respectively:

3 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) Fig. 1. Graphical representations of the data v t ¼ðv tint;v tmedþ centered at c k ¼ðc kint ; c kmed Þ with the number of clusters = 2. Fig. 2. Visualization of two quadrant axes built for two initial clusters that are centered at (c 1INT, c 1MED ) and (c 2INT, c 2MED ). v t ¼ðv tint ;v tmed Þ c k ¼ðc kint ; c kmed Þ where c kint and c kmed are defined by (4) and (5), respectively:! c kint ¼ 1 X n k c kmed ¼ 1 n k t2c k v tint! X t2c k v tmed ð2þ ð3þ ð4þ ð5þ In (2) (5), v tint is the intensity (INT) dimension of tth data, v tmed is the intensity median (MED) dimension of tth data, and n k is the number of pixels assigned to kth center. The graphical two-dimensional annotation of data with two obvious separated regions, along with the aforementioned parameters, is illustrated in Fig. 1. This dataset is segmented into two clusters (n c ¼ 2). In the proposed A2DKM algorithm, each region of the kth cluster is split into four quadrants (q = 1, 2, 3, 4) by setting the origin at (c kint, c kmed ), as shown in Fig. 2. The total Euclidean distance of all members in each quadrant of c k is calculated using: d kq ¼ X i2c kqkv i c kq k for q ¼ 1; 2; 3; 4 ð6þ where c kq is the qth quadrant of the kth cluster, and v i is the ith member of c kq. The quadrant with the largest total Euclidean distance is then determined and denoted as d kqmax. The total Euclidean distances of all four quadrants are compared with the d kqmax to fulfill the following condition:

4 910 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) d kqmax < 0:5 X4 q¼1 d kq! ð7þ If (7) is not fulfilled, then the kth center is split into two different clusters. The first cluster consists of all the members of the largest quadrant, while the other three quadrants form the second cluster. The center of each cluster is then recalculated based on (3) (5) because changes are necessary when members belong to newly-formed clusters. The processes of splitting the clusters into quadrants and forming new clusters are repeated until (7) is fulfilled. After (7) is fulfilled, no further formation of new clusters is necessary. The last number of clusters that was calculated is set as the final (or optimum) cluster number. This process could automatically determine the number of clusters; clustering starts with two clusters as the initial value because a data set must be partitioned into at least two groups with different cluster centers for the clustering analysis to have meaning [25]. 3. Illustration of implementation To illustrate the implementation of the A2DKM clustering algorithm, this study has applied the proposed algorithm on an image called Capsicum with the size of , as shown in Fig. 3a. The intensity median (for an N N window where N is an odd number that is set to 3 in this study) versus the intensity graph is plotted by the scatter graph shown in Fig. 3. Initially, the Capsicum image is clustered into two regions with the center values c 1 and c 2 set to (77,82) and (155,151), respectively. Using the proposed algorithm, the distance between each data point to all centers is calculated using a twodimensional (2D) Euclidean distance equation, which is then assigned to the closest center. The output of the assigning process is visualized using the pixel distribution graph in Fig. 3b. Four quadrants are subsequently built in each region, which are centered at c 1 and c 2. The data in each region belongs to one of the quadrants, as shown in Fig. 3c. The total 2D Euclidean distance for each quadrant to its center is then calculated. This method is used to decide the possibility of forming new clusters. In this example, Eq. (7) is not satisfied by centers, c 1 and c 2. The red 1 arrows in Fig. 3d show that d 14MAX > 0:5 P 4 q¼1 d 1q (because quadrant 4 of c 1 produces the maximum total Euclidean distance) for c 1, whereas d 22MAX > 0:5 P 4 q¼1 d 2q for c 2 (because quadrant 2 of c 2 produces the maximum total Euclidean distance). For c 1, quadrant 4 forms an individual cluster, whereas quadrants 1, 2, and 3 form the second cluster. For c 2, quadrant 2 forms one cluster, and the other three quadrants form the second cluster. Thus, the number of clusters increases from two to four, as depicted in Fig. 3d. The process is repeated until (7) is fulfilled. As shown in Fig. 3e, the final number of clusters obtained for Capsicum is 8. The resultant image obtained for the image labeled Capsicum (with n c = 8) is shown in Fig. 3f. In each graph in Fig. 3b e, different symbols represent different clusters. For the quadrant formation process, symbols colored with gray, red, blue, and green represent data belonging to quadrants 1, 2, 3, and 4, respectively. The red circles represent the initial centers, whereas the yellow circles represent the newly-formed centers. 4. Data analysis The objective of this study is to develop a clustering algorithm that is able to automatically segment any given image into its optimum segmentation regions without the initial determination of the number of clusters. The proposed algorithm is tested against synthetic images as the ground truth, wherein the number of clusters to be segmented is known. This verification is important to prove that the A2DKM algorithm is able to define the correct optimum number of clusters for an image. The method of determining the number of clusters for standard images is very subjective. Likewise, this determination is proven to be difficult. The ten synthetic images used in this study are images that were manually generated using Microsoft Ò Paint (version 5.1) in real colors. Furthermore, these synthetic images are converted into 8-bit gray scale images (with intervals from 0 to 255) using MATLAB Ò (version 7.5; R2007b). To analyze the performance of the A2DKM clustering algorithm in consumer-related applications (i.e., digital cameras, CCD cameras, and photo editing software), 100 gray-scale standard consumer images have been tested using the proposed algorithm. The same images are likewise processed using the conventional KM, FCM, and MKM clustering algorithms using the same number of clusters that was calculated using the A2DKM clustering algorithm. The performance of one of the latest clustering algorithms, namely, the AFKM is included in the comparison by using the same approach. The qualitative results (i.e., the results which are visible to the users) of these test images are crucial for any image processing tool. Thus, we have chosen to first qualitatively evaluate the images and then subsequently strengthen the findings by conducting a quantitative analysis. Qualitatively, high-quality segmented images are homogeneous and have uniform regions, with the minimum appearance of small isolated regions. In addition, adjacent regions of the segmented images should have significantly different values for uniform characteristics to produce a desirable clustering result [26]. In addition to the qualitative findings, four parameters are employed as the quantitative analysis benchmarks, namely, F(I), F 0 (I), Q(I), and the execution time t. 1 For interpretation of color in Fig. 3, the reader is referred to the web version of this article.

5 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) (b) Initialization of centers, c1 and c2 (a) (c) Quadrant formation of initial clusters (e) Quadrant formation of nc = 8 (d) Cluster updated from nc = 2 to nc = 4 (f) A2DKM segmentation result of image Capsicum Fig. 3. Implementation of the A2DKM algorithm. 911

6 912 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) F(I) was designed by Liu et al. in 1994 [27] as: 1 pffiffiffi X R e FðIÞ ¼ R pffiffiffiffi i 1000ðN MÞ A i i¼1 ð8þ where I is the segmented image, N M is the image size, R is the number of regions in the clustered image, A i is the area, and e i is the Euclidean distance between the gray-level color vectors of the pixels of the ith region and the color vector attributed to region i in the segmented image. Borsotti et al. [28] revised the F(I) function in 1998 and proposed an enhanced version of F(I), that is, F 0 (I) and Q(I): vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 ux FðIÞ ¼ t Max ½RðAÞŠ 1þ1=A XR e 2i pffiffiffiffi ð9þ 10; 000ðN MÞ A i A¼1 i¼1 1 pffiffiffi X R QðIÞ ¼ R 10; 000ðN MÞ i¼1 " e 2 i þ RðA # 2 iþ 1 þ log A i A i ð10þ where R(A) is the number of regions with the area A in (9), and Max is the area of the largest region in the segmented image. R(A i ) is the number of regions with an area equal to A i in (10). The three quantitative analysis functions are used to evaluate the segmentation results of an image, with as much correspondence as possible to the visual judgment. As observed from the functions, less homogeneous and uniform regions of segmented images with too many isolated small regions or noise yield higher values of F(I), F 0 (I), and Q(I). The F 0 (I) penalized the segmentation with too many small regions, whereas Q(I) penalized the segmentation with small regions and regions with large color errors. Thus, the difference between these three functions is the level of its resultant values that correspond to the visual judgment of segmented images. Although [28] proved that F 0 (I) and Q(I) closely correspond to the visual Table 1 Automated segmentation results using the proposed A2DKM for synthetic images. Image Original A2DKM Image Original A2DKM SYN1 SYN6 SYN2 Cluster no: 5 Cluster no: 5 Cluster no: 3 Cluster no: 3 SYN7 SYN3 Cluster no: 4 Cluster no: 4 Cluster no: 3 Cluster no: 3 SYN8 SYN4 Cluster no: 3 Cluster no: 3 Cluster no: 4 Cluster no: 4 SYN9 SYN5 Cluster no: 4 Cluster no: 4 Cluster no: 3 Cluster no: 3 SYN10 Cluster no: 3 Cluster no: 3 Cluster no: 3 Cluster no: 3

7 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) judgment, F(I) is still included in the quantitative analysis for a broader perspective on segmentation. In this analysis, the lower values of the three functions prove that the clustering algorithm produced more homogeneous, uniform, and smooth segmentation results. Execution time is used in this study as the fourth parameter in the quantitative analysis. The execution time analysis could be useful for evaluating the complexity of the proposed algorithm. A simpler algorithm requires shorter processing time, whereas a more complex algorithm has a longer processing time. This analysis penalizes the algorithms that require repetitive procedures. Moreover, we included memory consumption as one of the criteria for evaluation to further evaluate the clustering algorithms. An algorithm that has minimum computation consumption by optimizing its memory usage is desirable. 5. Results and discussion 5.1. Qualitative analysis The first part of the qualitative analysis is to prove that the proposed A2DKM clustering algorithm is able to give the optimum number of clusters for the clustering process. The proposed A2DKM is tested against synthetic images, and is able to accurately estimate the number of clusters that is suitable for each image. The A2DKM clustering processes for 10 synthetic images are summarized in Table 1, with the number of expected clusters and the resultant number of clusters from A2DKM. The corresponding results for ten synthetic images are presented in Table 1. Through this testing method, the proposed A2DKM is proven to be able to automatically and successfully determine the exact number of clusters of an image that is to be segmented for optimal segmentation quality. The objective of the qualitative analysis on standard consumer images is to virtually evaluate the appearances of the clustering results. The performance of the proposed A2DKM clustering algorithm is compared with that of the KM, FCM, MKM, and AFKM clustering algorithms. Out of the 100 gray-scale standard consumer images that were tested, three images (Lake, House, and Hut) are chosen for detailed qualitative evaluation. The number of clusters for Lake, House, and Hut images are 5, 7, and 4, respectively, as obtained by the proposed algorithm. These numbers of clusters are then used to set the number of clusters for the KM, FCM, MKM, and AFKM clustering algorithms. (a) (b) No. of regions = 4012, F(I) = , F (I) = 39.13, Q(I) = , t = 0.96 s (c) No. of regions = 3249, F(I) = , F (I) = 50.56, Q(I) = , t = 2.10 s (d) No. of regions = 3986, F(I) = , F (I) = 42.72, Q(I) = , t = 0.91 s (e) No. of regions = 4032, F(I) = , F (I) = 38.49, Q(I) = , t = 4.27 s (f) No. of regions = 2337, F(I) = , F (I) = 40.07, Q(I) = , t = 5.96 s Fig. 4. The Lake image. (a) Original image as well as the segmentation results using the (b) KM, (c) FCM, (d) MKM, (e) AFKM, and (f) A2DKM algorithms. The best results are in boldface.

8 914 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) The original images and the segmented results that were obtained using the KM, FCM, MKM, AFKM, and A2DKM clustering algorithms for the images labeled Lake, House, and Hut, respectively, are displayed in Figs The significant differences among the clustering algorithms in terms of their performance are highlighted by arrows. The Lake image in Fig. 4 is homogeneously segmented with smoother areas of the lake surface (without the wave pattern on its water surface) in the resultant image of the A2DKM clustering algorithm. The tree bark on the right corner and the tree leaves on the left corner of the image are likewise more homogeneous according to the image that was segmented using the A2DKM clustering algorithm as compared with those of the other clustering algorithms. All the other clustering algorithms insignificantly segment these areas, with several small isolated regions in the aforementioned areas. The proposed A2DKM clustering algorithm produces the most uniform and homogeneous wall area for the House image in Fig. 5 (i.e., the edges of bricks on the wall area are successfully removed). The number of small isolated regions in this area is much less as compared with those segmented using the KM, FCM, MKM, and AFKM clustering algorithms. Furthermore, the roof and its shadowy areas are more uniform in the resultant House image that was segmented using the proposed clustering algorithm than in the images segmented by the rest of the clustering algorithms, such that the insignificant details (i.e., edges of each piece of roof) and regions are successfully removed. Segmenting the Hut image using the proposed A2DKM produces a similar high level of performance, as qualitatively seen in Fig. 6. A big area of bushes and the roof of the hut in the image are successfully segmented into solid and less heterogeneous regions by the A2DKM clustering algorithm. By contrast, the KM, FCM, MKM, and AFKM clustering algorithms have segmented the region into many small regions, which are insignificant for image segmentation. The A2DKM clustering algorithm successfully segments the more homogeneous bush regions at the background by ignoring the different intensity levels of the leaves. The FCM clustering algorithm is able to segment the roof area into a single homogeneous region. However, the algorithm is unable to segment the bush regions into a more uniform region, thereby indicating a degree of inaccurate segmentation performance. Therefore, from the detailed analysis of the three standard consumer images, the proposed A2DKM clustering algorithm is able to automatically determine the optimal number of clusters for segmentation. In addition, the proposed A2DKM clustering algorithm qualitatively performs better as compared with the KM, FCM, MKM, and AFKM clustering algorithms by using the number of clusters which have been automatically obtained at the earlier stage of segmentation. Furthermore, this qualitative analysis proves that the local and spatial characteristics that were employed in the proposed A2DKM clustering algorithm are highly capable of reducing the segmentation of insignificant regions, thereby ensuring more homogeneous segmented regions. (a) (b) No. of regions = 1651, F(I) = , F (I) = 14.68, Q(I) = , t = 0.98 s (c) No. of regions = 1888, F(I) = , F (I) = 19.50, Q(I) = , t = 3.89 s (d) No. of regions = 1426, F(I) = , F (I) = 15.96, Q(I) = , t = 0.93 s (e) No. of regions = 1730, F(I) = , F (I) = 14.65, Q(I) = , t = 3.15 s (f) No. of regions = 925, F(I) = , F (I) = 13.90, Q(I) = , t = 3.04 s Fig. 5. The House image. (a) Original image as well as the segmentation results using the (b) KM, (c) FCM, (d) MKM, (e) AFKM, and (f) A2DKM algorithms. The best results are in boldface.

9 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) (a) (b) No. of regions = 2712, F(I) = , F (I) = , Q(I) = , t = 0.93 s (c) No. of regions = 3648, F(I) = , F (I) = , Q(I) = , t = 2.17 s (d) No. of regions = 2703, F(I) = , F (I) = , Q(I) = , t = 0.93 s (e) No. of regions = 2712, F(I) = , F (I) = 63.65, Q(I) = , t = 3.15 s (f) No. of regions = 1858, F(I) = , F (I) = 61.69, Q(I) = , t = 3.04 s Fig. 6. The Hut image. (a) Original image as well as the segmentation results using the (b) KM, (c) FCM, (d) MKM, (e) AFKM, and (f) A2DKM algorithms. The best results are in boldface Quantitative analysis The corresponding quantitative results for the Lake, House, and Hut images are presented below each image in Figs. 4 6, respectively. The best results are highlighted in bold. The quantitative analysis for the three selected images (i.e., Lake, House, and Hut) in those figures showed that the proposed A2DKM clustering algorithm gives the lowest readings of F(I) and F 0 (I) for two out of three images and gives the best readings of Q(I) for all three images. The AFKM clustering algorithm gives better F(I) and F 0 (I) values than the proposed A2DKM clustering algorithm for the Lake image, with a less than 2% difference in the results. This difference is comparable because the qualitative result of the proposed A2DKM clustering algorithm for Lake is much more homogeneous than that of the AFKM clustering algorithm. In addition, the smallest number of regions is obtained by the proposed A2DKM algorithm for all three images, thereby proving its ability to produce more homogeneous segmented regions with less unwanted and scattered small regions. The execution time analysis showed that the proposed A2DKM clustering algorithm takes a longer computational time than the KM, FCM, MKM, and AFKM clustering algorithms, with the maximum time interval of 5 s for the House image. This result is due to the fact that the A2DKM clustering algorithm is bound to be more complex compared to the other clustering algorithms to be able to automatically determine the number of optimal clusters for an image. Although the KM, FCM, MKM, and AFKM clustering algorithms have the shortest execution time, the number of clusters still needs to be predetermined by the user in these algorithms. To further verify the overall qualitative performance of the proposed A2DKM clustering algorithm as compared with the KM, FCM, MKM, and AFKM clustering algorithms, the qualitative readings of 100 gray-scale standard consumer images are collected. The average values for all quantitative tests are tabulated in Table 2. In two out of three qualitative tests (i.e., F(I) and Q(I)), the proposed A2DKM clustering algorithm produces the best results as compared with its KM, FCM, MKM, and AFKM clustering counterparts. From the observations regarding the execution time, the proposed A2DKM algorithm is not the best performer. However, the measurement of execution time t did not include the time it takes a user to define the number of clusters. In the implementation of the A2DKM clustering algorithm, the process of finding the number of clusters as well as the clustering of pixels is a continuous and repetitive process, until the optimum number of clusters is obtained. Thus, a longer processing time is inevitable because it has to include the time taken for the algorithm to obtain the number of clusters, as well as the time required for the actual clustering to occur. The ability to automatically determine the number of clusters by applying the A2DKM clustering algorithm can compensate for the time loss.

10 916 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) Table 2 Average quantitative results for 100 standard consumer images. The best results are made bold. Algorithm KM FCM MKM AFKM A2DKM Parameter F(I) (1.0e+3) F (I) (1.0e+2) Q(I) (1.0e+4) t (s) Memory requirement (MB) As highlighted in Table 2, the proposed A2DKM requires more memory consumption as compared with the conventional KM, FCM, and MKM clustering algorithms because of its continuous and repetitive process. However, the proposed A2DKM would require less memory consumption as compared with the AFKM algorithm. Thus, both the qualitative and quantitative analyses prove that the proposed A2DKM clustering algorithm has relatively better performance as compared with the KM, FCM, MKM, and AFKM clustering algorithms. Despite its complexity and higher memory requirement, the results prove that the A2DKM algorithm produces better segmented images when compared with other clustering algorithms. The proposed A2DKM requires an extra processing time which is still within the acceptable range for its reliable application. 6. Conclusion A novel unsupervised clustering algorithm, namely, the Automated Two-Dimensional K-Means (A2DKM) has been proposed and introduced in this study. To evaluate its performance for consumer digital imaging applications, the A2DKM clustering algorithm is qualitatively and quantitatively tested using synthetics and standard consumer images. Based on the results, the proposed A2DKM clustering algorithm is able to automatically determine the optimal number of clusters for image segmentation. The A2DKM algorithm qualitatively and quantitatively demonstrated better performance as compared with the conventional KM, FCM, and MKM clustering algorithms. By further comparing its performance against one of the latest clustering algorithms (the AFKM clustering algorithm), the A2DKM algorithm shows similar results. Qualitatively, the images produced by the proposed A2DKM algorithm are more homogeneous and uniform, with smoother segmented regions. Quantitatively, the A2DKM clustering algorithm produces lower F(I) and Q(I) values and comparable readings of F 0 (I), which are desired in image segmentation. Execution times of the proposed algorithm are still comparable to those of the KM, FCM, MKM, and AFKM clustering algorithms because the A2DKM clustering algorithm completely omits the need to declare the number of clusters during the initial stage of processing. Finally, the memory requirement of A2DKM is proven to be smaller than the amount required by the AFKM algorithm. Therefore, the proposed novel A2DKM clustering algorithm is able to automatically determine the optimal number of clusters for segmentation. The A2DKM clustering algorithm performs better than the KM, FCM, MKM, and the latest AFKM clustering algorithms in terms of image quality, which is further proven in the quantitative analysis. Given its excellent capabilities, despite applying it to the hard clustering algorithm (i.e. KM algorithm) the concept employed in the proposed A2DKM could be further applied in the soft clustering algorithm such as the FCM. Furthermore, the proposed A2DKM algorithm could be applied to color images, where the robustness of the algorithm can be increased, thereby further supporting its high efficiency as an unsupervised segmentation technique. Acknowledgement This work was partially supported by the Universiti Sains Malaysia Short Term Grant entitled Fuzzy Logic Based Segmentation Technique for Determination of Breast Tumor on Mammogram Image. References [1] Brosnan T, Sun D-W. Inspection and grading of agricultural and food products by computer vision systems a review. Comput Electron Agric 2002;36(2 3): [2] Leemans V, Magein H, Destain M-F. Defects segmentation on Golden Delicious apples by using colour machine vision. Comput Electron Agric 1998;20(2):117 /130. [3] Liming X, Yanchao Z. Automated strawberry grading system based on image processing. Comput Electron Agric 2010;71(1):S32 9. [5] Xu B, Zhuang Y, Tang H, Zhang L. Object-based multilevel contrast stretching method for image enhancement. IEEE Trans Consum Electron 2010;56(3): [6] Lee E, Kang W, Kim S, Paik J. Color shift model-based image enhancement for digital multifocusing based on a multiple color-filter aperture camera. IEEE Trans Consum Electron 2010;56(2): [7] Bhukhanwala SA, Ramabadran TV. Automated global enhancement of digitized photographs. IEEE Trans Consum Electron 1994;40(1):1 10. [8] Lee JW, Park R-H, Chang S. Local tone mapping using the K-means algorithm and automatic gamma setting. IEEE Trans Consum Electron 2011;57(1): [9] Warfield SK, Michael K, Jolesz FA, Ron K. Adaptive, template moderated, spatially varying statistical classification. Med Image Anal 2000;4(1): [10] Cheriet M, Said JN, Suen CY. A recursive thresholding technique for image segmentation. IEEE Trans Image Process 1998;7(6): [11] Seunghwan Y, Rae-Hong P. Red-eye detection and correction using inpainting in digital photographs. IEEE Trans Consum Electron 2009;55(3):

11 I.A. Yusoff et al. / Computers and Electrical Engineering 39 (2013) [12] Xiaopeng Y, Deyi Z, Yan Z. A new clustering algorithm based on distance and density. In: Proceedings of international conference on services systems and services management, ICSSSM 05, vol. 2; p [13] Dave RN, Krishnapuram R. Robust clustering methods: a unified view. IEEE Trans Fuzzy Syst 1997;5(2): [14] Verma NK, Roy A, Vasikarla S. Medical image segmentation using improved mountain clustering technique version-2. In: 7th International conference on information technology: new generations (ITNG); p [15] Wang Z. Comparison of four kinds of fuzzy C-means clustering methods. In: Third international symposium on information processing (ISIP); p [16] Mashor MY. Hybrid training algorithm for RBF network. Int J Comput, Internet Manage 2000;8(2): [17] Sulaiman SN, Isa NAM. Adaptive fuzzy-k-means clustering algorithm for image segmentation. IEEE Trans Consum Electron 2010;56(4): [18] Elsayad AM. Completely unsupervised image segmentation using wavelet analysis and Gustafson-Kessel clustering. In: 5th International multiconference on systems, signals and devices, IEEE SSD; p [19] Sahaphong S, Hiransakolwong N. Unsupervised image segmentation using automated fuzzy c-means. In: 7th IEEE International conference on computer and information technology, CIT; p [20] Feng J, Mingjing L, Hong-Jiang Z, Bo Z. Unsupervised image segmentation using local homogeneity analysis. In: Proceedings of the 2003 international symposium on circuits and systems, ISCAS 03, vol. 2; p. II [21] Hu LP, Liu HW, Yin KY, Wu SJ. Two-dimensional clustering-based discriminant analysis for SAR ATR. In: 8th International symposium on antennas, propagation and EM theory, ISAPE; 2 5 November p [22] Chen S, Zhang D. Robust image segmentation using FCM with spatial constraints based on new kernel-induced distance measure. IEEE Trans Syst Man Cybern B: Cybern 2004;34(4): [23] Chuang KS, Tzeng HL, Chen S, Wu J, Chen TJ. Fuzzy c-means clustering with spatial information for image segmentation. Comput Med Imag Graph 2006:9 15. [24] Li L, Gong R, Chen W. Gray level image thresholding based on fisher linear projection of two-dimensional histogram. Pattern Recogn 1997;30: [25] Davies DL, Bouldin DW. A cluster separation measure. IEEE Trans Pattern Anal Mach Intell 1979;PAMI-1(2): [26] Haralick RH, Shapiro LG. Image segmentations techniques. Comput Vis Graph Image Process 1985;29: [27] Liu J, Yang YH. Multiresolution color image segmentation. IEEE Trans Pattern Anal Mach Intell 1994;16(7): [28] Borsotti M, Campadelli P, Schettini R. Quantitative evaluation of color image segmentation results. Pattern Recogn Lett 1998;19(8): Intan Aidha Yusoff obtained her BEng (Hons) in Electronics Engineering from the Universiti Sains Malaysia (USM) in She is currently pursuing her MSc degree in Electrical and Electronics Engineering in USM while attached to the Imaging and Intelligent System Research Team (ISRT), School of Electrical and Electronic Engineering (SEEE), USM, Engineering Campus in Nibong Tebal, Penang, Malaysia. Nor Ashidi Mat Isa received his BEng (in 1999) and PhD (in 2003) in Electrical and Electronics Engineering from USM. He is currently an Associate Professor at the SEEE, USM. His research interests include intelligent systems, image processing, biomedical engineering, and algorithms. He has published numerous journals, book chapters, and proceedings at both the national and international arena. Khairunnisa Hasikin obtained her BEng in Electrical Engineering and MEngSc from the University of Malaya (UM) in 2007 and 2010, respectively. She is currently pursuing her PhD in video/image processing while attached to the ISRT, SEEE, USM, Engineering Campus in Nibong Tebal, Penang, Malaysia.

Two-Dimensional Clustering Algorithms for Image Segmentation

Two-Dimensional Clustering Algorithms for Image Segmentation Two-Dimensional Clustering Algorithms for Segmentation INTAN AIDHA YUSOFF 1, NOR ASHIDI MAT ISA 2 Imaging and Intelligent System Research Team (ISRT) Universiti Sains Malaysia Engineering Campus Nibong

More information

Image Segmentation Using Two Weighted Variable Fuzzy K Means

Image Segmentation Using Two Weighted Variable Fuzzy K Means Image Segmentation Using Two Weighted Variable Fuzzy K Means S. Suganya Rose Margaret CMS College of Science and Commerce CMS College of Science and Commerce Coimbatore, Tamil Nadu, India Coimbatore, Tamil

More information

Texture Image Segmentation using FCM

Texture Image Segmentation using FCM Proceedings of 2012 4th International Conference on Machine Learning and Computing IPCSIT vol. 25 (2012) (2012) IACSIT Press, Singapore Texture Image Segmentation using FCM Kanchan S. Deshmukh + M.G.M

More information

CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION

CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION CHAPTER 6 MODIFIED FUZZY TECHNIQUES BASED IMAGE SEGMENTATION 6.1 INTRODUCTION Fuzzy logic based computational techniques are becoming increasingly important in the medical image analysis arena. The significant

More information

Color Image Segmentation

Color Image Segmentation Color Image Segmentation Yining Deng, B. S. Manjunath and Hyundoo Shin* Department of Electrical and Computer Engineering University of California, Santa Barbara, CA 93106-9560 *Samsung Electronics Inc.

More information

Improving the efficiency of Medical Image Segmentation based on Histogram Analysis

Improving the efficiency of Medical Image Segmentation based on Histogram Analysis Advances in Computational Sciences and Technology ISSN 0973-6107 Volume 10, Number 1 (2017) pp. 91-101 Research India Publications http://www.ripublication.com Improving the efficiency of Medical Image

More information

Color-Texture Segmentation of Medical Images Based on Local Contrast Information

Color-Texture Segmentation of Medical Images Based on Local Contrast Information Color-Texture Segmentation of Medical Images Based on Local Contrast Information Yu-Chou Chang Department of ECEn, Brigham Young University, Provo, Utah, 84602 USA ycchang@et.byu.edu Dah-Jye Lee Department

More information

Available Online through

Available Online through Available Online through www.ijptonline.com ISSN: 0975-766X CODEN: IJPTFI Research Article ANALYSIS OF CT LIVER IMAGES FOR TUMOUR DIAGNOSIS BASED ON CLUSTERING TECHNIQUE AND TEXTURE FEATURES M.Krithika

More information

A Study on Similarity Computations in Template Matching Technique for Identity Verification

A Study on Similarity Computations in Template Matching Technique for Identity Verification A Study on Similarity Computations in Template Matching Technique for Identity Verification Lam, S. K., Yeong, C. Y., Yew, C. T., Chai, W. S., Suandi, S. A. Intelligent Biometric Group, School of Electrical

More information

Improved Version of Kernelized Fuzzy C-Means using Credibility

Improved Version of Kernelized Fuzzy C-Means using Credibility 50 Improved Version of Kernelized Fuzzy C-Means using Credibility Prabhjot Kaur Maharaja Surajmal Institute of Technology (MSIT) New Delhi, 110058, INDIA Abstract - Fuzzy c-means is a clustering algorithm

More information

IJSER. Real Time Object Visual Inspection Based On Template Matching Using FPGA

IJSER. Real Time Object Visual Inspection Based On Template Matching Using FPGA International Journal of Scientific & Engineering Research, Volume 4, Issue 8, August-2013 823 Real Time Object Visual Inspection Based On Template Matching Using FPGA GURURAJ.BANAKAR Electronics & Communications

More information

Improvement of Contrast Enhancement Technique for Cervical Cell of Pap Smear Images by Reducing the Effect of Unwanted Background Information

Improvement of Contrast Enhancement Technique for Cervical Cell of Pap Smear Images by Reducing the Effect of Unwanted Background Information Improvement of Contrast Enhancement Technique for Cervical Cell of Pap Smear Images by Reducing the Effect of Unwanted Background Information * Nor Ashidi Mat Isa, Nazahah Mustafa, Kamal Zuhairi Zamli,

More information

Real Time Object Visual Inspection Based On Template Matching Using FPGA

Real Time Object Visual Inspection Based On Template Matching Using FPGA Real Time Object Visual Inspection Based On Template Matching Using FPGA SANDHYA P Asst Professor in MTIET Palamaner, India OMKAR NAIDU V Asst Professor in MTIET Palamaner, India Abstract In similarity

More information

Denoising of Noisy MRI Brain Image by Using Switching-based Clustering Algorithm

Denoising of Noisy MRI Brain Image by Using Switching-based Clustering Algorithm Denoising of Noisy MRI Brain Image by Using -based Clustering Algorithm Siti Noraini Sulaiman, Siti Mastura Che Isha, Iza Sazanita Isa, Norhazimi Hamzah Abstract Magnetic Resonance Image is one of technologies

More information

Image Segmentation Based on Watershed and Edge Detection Techniques

Image Segmentation Based on Watershed and Edge Detection Techniques 0 The International Arab Journal of Information Technology, Vol., No., April 00 Image Segmentation Based on Watershed and Edge Detection Techniques Nassir Salman Computer Science Department, Zarqa Private

More information

Available online Journal of Scientific and Engineering Research, 2019, 6(1): Research Article

Available online   Journal of Scientific and Engineering Research, 2019, 6(1): Research Article Available online www.jsaer.com, 2019, 6(1):193-197 Research Article ISSN: 2394-2630 CODEN(USA): JSERBR An Enhanced Application of Fuzzy C-Mean Algorithm in Image Segmentation Process BAAH Barida 1, ITUMA

More information

Automatic thresholding for defect detection

Automatic thresholding for defect detection Pattern Recognition Letters xxx (2006) xxx xxx www.elsevier.com/locate/patrec Automatic thresholding for defect detection Hui-Fuang Ng * Department of Computer Science and Information Engineering, Asia

More information

A Survey on Image Segmentation Using Clustering Techniques

A Survey on Image Segmentation Using Clustering Techniques A Survey on Image Segmentation Using Clustering Techniques Preeti 1, Assistant Professor Kompal Ahuja 2 1,2 DCRUST, Murthal, Haryana (INDIA) Abstract: Image is information which has to be processed effectively.

More information

Efficient Image Compression of Medical Images Using the Wavelet Transform and Fuzzy c-means Clustering on Regions of Interest.

Efficient Image Compression of Medical Images Using the Wavelet Transform and Fuzzy c-means Clustering on Regions of Interest. Efficient Image Compression of Medical Images Using the Wavelet Transform and Fuzzy c-means Clustering on Regions of Interest. D.A. Karras, S.A. Karkanis and D. E. Maroulis University of Piraeus, Dept.

More information

A Quantitative Approach for Textural Image Segmentation with Median Filter

A Quantitative Approach for Textural Image Segmentation with Median Filter International Journal of Advancements in Research & Technology, Volume 2, Issue 4, April-2013 1 179 A Quantitative Approach for Textural Image Segmentation with Median Filter Dr. D. Pugazhenthi 1, Priya

More information

NCC 2009, January 16-18, IIT Guwahati 267

NCC 2009, January 16-18, IIT Guwahati 267 NCC 2009, January 6-8, IIT Guwahati 267 Unsupervised texture segmentation based on Hadamard transform Tathagata Ray, Pranab Kumar Dutta Department Of Electrical Engineering Indian Institute of Technology

More information

Iterative Removing Salt and Pepper Noise based on Neighbourhood Information

Iterative Removing Salt and Pepper Noise based on Neighbourhood Information Iterative Removing Salt and Pepper Noise based on Neighbourhood Information Liu Chun College of Computer Science and Information Technology Daqing Normal University Daqing, China Sun Bishen Twenty-seventh

More information

Colour Segmentation-based Computation of Dense Optical Flow with Application to Video Object Segmentation

Colour Segmentation-based Computation of Dense Optical Flow with Application to Video Object Segmentation ÖGAI Journal 24/1 11 Colour Segmentation-based Computation of Dense Optical Flow with Application to Video Object Segmentation Michael Bleyer, Margrit Gelautz, Christoph Rhemann Vienna University of Technology

More information

Image Segmentation Via Iterative Geodesic Averaging

Image Segmentation Via Iterative Geodesic Averaging Image Segmentation Via Iterative Geodesic Averaging Asmaa Hosni, Michael Bleyer and Margrit Gelautz Institute for Software Technology and Interactive Systems, Vienna University of Technology Favoritenstr.

More information

A New iterative triclass thresholding technique for Image Segmentation

A New iterative triclass thresholding technique for Image Segmentation A New iterative triclass thresholding technique for Image Segmentation M.M.Raghavendra Asst Prof, Department of ECE Brindavan Institute of Technology & Science Kurnool, India E-mail: mmraghavendraece@gmail.com

More information

COLOR image segmentation is a method of assigning

COLOR image segmentation is a method of assigning Proceedings of the Federated Conference on Computer Science and Information Systems pp. 1049 1054 DOI: 10.15439/2015F222 ACSIS, Vol. 5 Applying fuzzy clustering method to color image segmentation Omer

More information

v = (V,,v2,..., v, ) and the set of arc E c v X v, and a

v = (V,,v2,..., v, ) and the set of arc E c v X v, and a Proceedings of the First International Conference on Machine Learning and Cybernetics, Beijing, 4-5 November 2002 TIME-VARIABLE-PARAMETRIC RELAXATION LABELING AND ITS APPLICATION IN TEXTURE SEGMENTATION

More information

A Laplacian Based Novel Approach to Efficient Text Localization in Grayscale Images

A Laplacian Based Novel Approach to Efficient Text Localization in Grayscale Images A Laplacian Based Novel Approach to Efficient Text Localization in Grayscale Images Karthik Ram K.V & Mahantesh K Department of Electronics and Communication Engineering, SJB Institute of Technology, Bangalore,

More information

A reversible data hiding based on adaptive prediction technique and histogram shifting

A reversible data hiding based on adaptive prediction technique and histogram shifting A reversible data hiding based on adaptive prediction technique and histogram shifting Rui Liu, Rongrong Ni, Yao Zhao Institute of Information Science Beijing Jiaotong University E-mail: rrni@bjtu.edu.cn

More information

A MORPHOLOGY-BASED FILTER STRUCTURE FOR EDGE-ENHANCING SMOOTHING

A MORPHOLOGY-BASED FILTER STRUCTURE FOR EDGE-ENHANCING SMOOTHING Proceedings of the 1994 IEEE International Conference on Image Processing (ICIP-94), pp. 530-534. (Austin, Texas, 13-16 November 1994.) A MORPHOLOGY-BASED FILTER STRUCTURE FOR EDGE-ENHANCING SMOOTHING

More information

Reversible Image Data Hiding with Local Adaptive Contrast Enhancement

Reversible Image Data Hiding with Local Adaptive Contrast Enhancement Reversible Image Data Hiding with Local Adaptive Contrast Enhancement Ruiqi Jiang, Weiming Zhang, Jiajia Xu, Nenghai Yu and Xiaocheng Hu Abstract Recently, a novel reversible data hiding scheme is proposed

More information

Non-Linear Masking based Contrast Enhancement via Illumination Estimation

Non-Linear Masking based Contrast Enhancement via Illumination Estimation https://doi.org/10.2352/issn.2470-1173.2018.13.ipas-389 2018, Society for Imaging Science and Technology Non-Linear Masking based Contrast Enhancement via Illumination Estimation Soonyoung Hong, Minsub

More information

COLOR BASED REMOTE SENSING IMAGE SEGMENTATION USING FUZZY C-MEANS AND IMPROVED SOBEL EDGE DETECTION ALGORITHM

COLOR BASED REMOTE SENSING IMAGE SEGMENTATION USING FUZZY C-MEANS AND IMPROVED SOBEL EDGE DETECTION ALGORITHM COLOR BASED REMOTE SENSING IMAGE SEGMENTATION USING FUZZY C-MEANS AND IMPROVED SOBEL EDGE DETECTION ALGORITHM Ms. B.SasiPrabha, Mrs.R.uma, MCA,M.Phil,M.Ed, Research scholar, Asst. professor, Department

More information

IRIS SEGMENTATION OF NON-IDEAL IMAGES

IRIS SEGMENTATION OF NON-IDEAL IMAGES IRIS SEGMENTATION OF NON-IDEAL IMAGES William S. Weld St. Lawrence University Computer Science Department Canton, NY 13617 Xiaojun Qi, Ph.D Utah State University Computer Science Department Logan, UT 84322

More information

Moving Object Segmentation Method Based on Motion Information Classification by X-means and Spatial Region Segmentation

Moving Object Segmentation Method Based on Motion Information Classification by X-means and Spatial Region Segmentation IJCSNS International Journal of Computer Science and Network Security, VOL.13 No.11, November 2013 1 Moving Object Segmentation Method Based on Motion Information Classification by X-means and Spatial

More information

High Resolution Remote Sensing Image Classification based on SVM and FCM Qin LI a, Wenxing BAO b, Xing LI c, Bin LI d

High Resolution Remote Sensing Image Classification based on SVM and FCM Qin LI a, Wenxing BAO b, Xing LI c, Bin LI d 2nd International Conference on Electrical, Computer Engineering and Electronics (ICECEE 2015) High Resolution Remote Sensing Image Classification based on SVM and FCM Qin LI a, Wenxing BAO b, Xing LI

More information

Novel Intuitionistic Fuzzy C-Means Clustering for Linearly and Nonlinearly Separable Data

Novel Intuitionistic Fuzzy C-Means Clustering for Linearly and Nonlinearly Separable Data Novel Intuitionistic Fuzzy C-Means Clustering for Linearly and Nonlinearly Separable Data PRABHJOT KAUR DR. A. K. SONI DR. ANJANA GOSAIN Department of IT, MSIT Department of Computers University School

More information

AN EFFICIENT BINARIZATION TECHNIQUE FOR FINGERPRINT IMAGES S. B. SRIDEVI M.Tech., Department of ECE

AN EFFICIENT BINARIZATION TECHNIQUE FOR FINGERPRINT IMAGES S. B. SRIDEVI M.Tech., Department of ECE AN EFFICIENT BINARIZATION TECHNIQUE FOR FINGERPRINT IMAGES S. B. SRIDEVI M.Tech., Department of ECE sbsridevi89@gmail.com 287 ABSTRACT Fingerprint identification is the most prominent method of biometric

More information

Change detection using joint intensity histogram

Change detection using joint intensity histogram Change detection using joint intensity histogram Yasuyo Kita National Institute of Advanced Industrial Science and Technology (AIST) Information Technology Research Institute AIST Tsukuba Central 2, 1-1-1

More information

Color-Based Classification of Natural Rock Images Using Classifier Combinations

Color-Based Classification of Natural Rock Images Using Classifier Combinations Color-Based Classification of Natural Rock Images Using Classifier Combinations Leena Lepistö, Iivari Kunttu, and Ari Visa Tampere University of Technology, Institute of Signal Processing, P.O. Box 553,

More information

Video Inter-frame Forgery Identification Based on Optical Flow Consistency

Video Inter-frame Forgery Identification Based on Optical Flow Consistency Sensors & Transducers 24 by IFSA Publishing, S. L. http://www.sensorsportal.com Video Inter-frame Forgery Identification Based on Optical Flow Consistency Qi Wang, Zhaohong Li, Zhenzhen Zhang, Qinglong

More information

Fuzzy C-means Clustering with Temporal-based Membership Function

Fuzzy C-means Clustering with Temporal-based Membership Function Indian Journal of Science and Technology, Vol (S()), DOI:./ijst//viS/, December ISSN (Print) : - ISSN (Online) : - Fuzzy C-means Clustering with Temporal-based Membership Function Aseel Mousa * and Yuhanis

More information

Texture recognition of medical images with the ICM method

Texture recognition of medical images with the ICM method Nuclear Instruments and Methods in Physics Research A 525 (2004) 387 391 Texture recognition of medical images with the ICM method Jason M. Kinser*, Guisong Wang George Mason University, Institute for

More information

Chapter 7 UNSUPERVISED LEARNING TECHNIQUES FOR MAMMOGRAM CLASSIFICATION

Chapter 7 UNSUPERVISED LEARNING TECHNIQUES FOR MAMMOGRAM CLASSIFICATION UNSUPERVISED LEARNING TECHNIQUES FOR MAMMOGRAM CLASSIFICATION Supervised and unsupervised learning are the two prominent machine learning algorithms used in pattern recognition and classification. In this

More information

HARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION

HARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION HARD, SOFT AND FUZZY C-MEANS CLUSTERING TECHNIQUES FOR TEXT CLASSIFICATION 1 M.S.Rekha, 2 S.G.Nawaz 1 PG SCALOR, CSE, SRI KRISHNADEVARAYA ENGINEERING COLLEGE, GOOTY 2 ASSOCIATE PROFESSOR, SRI KRISHNADEVARAYA

More information

Image Segmentation Techniques

Image Segmentation Techniques A Study On Image Segmentation Techniques Palwinder Singh 1, Amarbir Singh 2 1,2 Department of Computer Science, GNDU Amritsar Abstract Image segmentation is very important step of image analysis which

More information

Cursive Handwriting Recognition System Using Feature Extraction and Artificial Neural Network

Cursive Handwriting Recognition System Using Feature Extraction and Artificial Neural Network Cursive Handwriting Recognition System Using Feature Extraction and Artificial Neural Network Utkarsh Dwivedi 1, Pranjal Rajput 2, Manish Kumar Sharma 3 1UG Scholar, Dept. of CSE, GCET, Greater Noida,

More information

Optik 124 (2013) Contents lists available at SciVerse ScienceDirect. Optik. jou rnal homepage:

Optik 124 (2013) Contents lists available at SciVerse ScienceDirect. Optik. jou rnal homepage: Optik 14 013 45 431 Contents lists available at SciVerse ScienceDirect Optik jou rnal homepage: www.elsevier.de/ijleo Range Limited Bi-Histogram Equalization for image contrast enhancement Chao Zuo, Qian

More information

Image Segmentation for Image Object Extraction

Image Segmentation for Image Object Extraction Image Segmentation for Image Object Extraction Rohit Kamble, Keshav Kaul # Computer Department, Vishwakarma Institute of Information Technology, Pune kamble.rohit@hotmail.com, kaul.keshav@gmail.com ABSTRACT

More information

A Fuzzy C-means Clustering Algorithm Based on Pseudo-nearest-neighbor Intervals for Incomplete Data

A Fuzzy C-means Clustering Algorithm Based on Pseudo-nearest-neighbor Intervals for Incomplete Data Journal of Computational Information Systems 11: 6 (2015) 2139 2146 Available at http://www.jofcis.com A Fuzzy C-means Clustering Algorithm Based on Pseudo-nearest-neighbor Intervals for Incomplete Data

More information

Color Local Texture Features Based Face Recognition

Color Local Texture Features Based Face Recognition Color Local Texture Features Based Face Recognition Priyanka V. Bankar Department of Electronics and Communication Engineering SKN Sinhgad College of Engineering, Korti, Pandharpur, Maharashtra, India

More information

Understanding Clustering Supervising the unsupervised

Understanding Clustering Supervising the unsupervised Understanding Clustering Supervising the unsupervised Janu Verma IBM T.J. Watson Research Center, New York http://jverma.github.io/ jverma@us.ibm.com @januverma Clustering Grouping together similar data

More information

CHAPTER 5 CLUSTER VALIDATION TECHNIQUES

CHAPTER 5 CLUSTER VALIDATION TECHNIQUES 120 CHAPTER 5 CLUSTER VALIDATION TECHNIQUES 5.1 INTRODUCTION Prediction of correct number of clusters is a fundamental problem in unsupervised classification techniques. Many clustering techniques require

More information

A Miniature-Based Image Retrieval System

A Miniature-Based Image Retrieval System A Miniature-Based Image Retrieval System Md. Saiful Islam 1 and Md. Haider Ali 2 Institute of Information Technology 1, Dept. of Computer Science and Engineering 2, University of Dhaka 1, 2, Dhaka-1000,

More information

CHAPTER 1 INTRODUCTION

CHAPTER 1 INTRODUCTION CHAPTER 1 INTRODUCTION 1.1 Introduction Pattern recognition is a set of mathematical, statistical and heuristic techniques used in executing `man-like' tasks on computers. Pattern recognition plays an

More information

A Hybrid Approach to CAM-Based Longest Prefix Matching for IP Route Lookup

A Hybrid Approach to CAM-Based Longest Prefix Matching for IP Route Lookup A Hybrid Approach to CAM-Based Longest Prefix Matching for IP Route Lookup Yan Sun and Min Sik Kim School of Electrical Engineering and Computer Science Washington State University Pullman, Washington

More information

BRIEF Features for Texture Segmentation

BRIEF Features for Texture Segmentation BRIEF Features for Texture Segmentation Suraya Mohammad 1, Tim Morris 2 1 Communication Technology Section, Universiti Kuala Lumpur - British Malaysian Institute, Gombak, Selangor, Malaysia 2 School of

More information

Structural Analysis of Aerial Photographs (HB47 Computer Vision: Assignment)

Structural Analysis of Aerial Photographs (HB47 Computer Vision: Assignment) Structural Analysis of Aerial Photographs (HB47 Computer Vision: Assignment) Xiaodong Lu, Jin Yu, Yajie Li Master in Artificial Intelligence May 2004 Table of Contents 1 Introduction... 1 2 Edge-Preserving

More information

DYNAMIC BACKGROUND SUBTRACTION BASED ON SPATIAL EXTENDED CENTER-SYMMETRIC LOCAL BINARY PATTERN. Gengjian Xue, Jun Sun, Li Song

DYNAMIC BACKGROUND SUBTRACTION BASED ON SPATIAL EXTENDED CENTER-SYMMETRIC LOCAL BINARY PATTERN. Gengjian Xue, Jun Sun, Li Song DYNAMIC BACKGROUND SUBTRACTION BASED ON SPATIAL EXTENDED CENTER-SYMMETRIC LOCAL BINARY PATTERN Gengjian Xue, Jun Sun, Li Song Institute of Image Communication and Information Processing, Shanghai Jiao

More information

Time Stamp Detection and Recognition in Video Frames

Time Stamp Detection and Recognition in Video Frames Time Stamp Detection and Recognition in Video Frames Nongluk Covavisaruch and Chetsada Saengpanit Department of Computer Engineering, Chulalongkorn University, Bangkok 10330, Thailand E-mail: nongluk.c@chula.ac.th

More information

An indirect tire identification method based on a two-layered fuzzy scheme

An indirect tire identification method based on a two-layered fuzzy scheme Journal of Intelligent & Fuzzy Systems 29 (2015) 2795 2800 DOI:10.3233/IFS-151984 IOS Press 2795 An indirect tire identification method based on a two-layered fuzzy scheme Dailin Zhang, Dengming Zhang,

More information

An Effective Denoising Method for Images Contaminated with Mixed Noise Based on Adaptive Median Filtering and Wavelet Threshold Denoising

An Effective Denoising Method for Images Contaminated with Mixed Noise Based on Adaptive Median Filtering and Wavelet Threshold Denoising J Inf Process Syst, Vol.14, No.2, pp.539~551, April 2018 https://doi.org/10.3745/jips.02.0083 ISSN 1976-913X (Print) ISSN 2092-805X (Electronic) An Effective Denoising Method for Images Contaminated with

More information

Norbert Schuff VA Medical Center and UCSF

Norbert Schuff VA Medical Center and UCSF Norbert Schuff Medical Center and UCSF Norbert.schuff@ucsf.edu Medical Imaging Informatics N.Schuff Course # 170.03 Slide 1/67 Objective Learn the principle segmentation techniques Understand the role

More information

A Study on various Histogram Equalization Techniques to Preserve the Brightness for Gray Scale and Color Images

A Study on various Histogram Equalization Techniques to Preserve the Brightness for Gray Scale and Color Images A Study on various Histogram Equalization Techniques to Preserve the Brightness for Gray Scale Color Images Babu P Balasubramanian.K Vol., 8 Abstract:-Histogram equalization (HE) wors well on singlechannel

More information

Image Segmentation Based on. Modified Tsallis Entropy

Image Segmentation Based on. Modified Tsallis Entropy Contemporary Engineering Sciences, Vol. 7, 2014, no. 11, 523-529 HIKARI Ltd, www.m-hikari.com http://dx.doi.org/10.12988/ces.2014.4439 Image Segmentation Based on Modified Tsallis Entropy V. Vaithiyanathan

More information

Biometrics Technology: Image Processing & Pattern Recognition (by Dr. Dickson Tong)

Biometrics Technology: Image Processing & Pattern Recognition (by Dr. Dickson Tong) Biometrics Technology: Image Processing & Pattern Recognition (by Dr. Dickson Tong) References: [1] http://homepages.inf.ed.ac.uk/rbf/hipr2/index.htm [2] http://www.cs.wisc.edu/~dyer/cs540/notes/vision.html

More information

CHAPTER 4 DETECTION OF DISEASES IN PLANT LEAF USING IMAGE SEGMENTATION

CHAPTER 4 DETECTION OF DISEASES IN PLANT LEAF USING IMAGE SEGMENTATION CHAPTER 4 DETECTION OF DISEASES IN PLANT LEAF USING IMAGE SEGMENTATION 4.1. Introduction Indian economy is highly dependent of agricultural productivity. Therefore, in field of agriculture, detection of

More information

HYBRID CENTER-SYMMETRIC LOCAL PATTERN FOR DYNAMIC BACKGROUND SUBTRACTION. Gengjian Xue, Li Song, Jun Sun, Meng Wu

HYBRID CENTER-SYMMETRIC LOCAL PATTERN FOR DYNAMIC BACKGROUND SUBTRACTION. Gengjian Xue, Li Song, Jun Sun, Meng Wu HYBRID CENTER-SYMMETRIC LOCAL PATTERN FOR DYNAMIC BACKGROUND SUBTRACTION Gengjian Xue, Li Song, Jun Sun, Meng Wu Institute of Image Communication and Information Processing, Shanghai Jiao Tong University,

More information

International Journal of Modern Engineering and Research Technology

International Journal of Modern Engineering and Research Technology Volume 4, Issue 3, July 2017 ISSN: 2348-8565 (Online) International Journal of Modern Engineering and Research Technology Website: http://www.ijmert.org Email: editor.ijmert@gmail.com A Novel Approach

More information

Study on fabric density identification based on binary feature matrix

Study on fabric density identification based on binary feature matrix 153 Study on fabric density identification based on binary feature matrix Xiuchen Wang 1,2 Xiaojiu Li 2 Zhe Liu 1 1 Zhongyuan University of Technology Zhengzhou, China 2Tianjin Polytechnic University Tianjin,

More information

A Survey of Light Source Detection Methods

A Survey of Light Source Detection Methods A Survey of Light Source Detection Methods Nathan Funk University of Alberta Mini-Project for CMPUT 603 November 30, 2003 Abstract This paper provides an overview of the most prominent techniques for light

More information

Classification. Vladimir Curic. Centre for Image Analysis Swedish University of Agricultural Sciences Uppsala University

Classification. Vladimir Curic. Centre for Image Analysis Swedish University of Agricultural Sciences Uppsala University Classification Vladimir Curic Centre for Image Analysis Swedish University of Agricultural Sciences Uppsala University Outline An overview on classification Basics of classification How to choose appropriate

More information

Automatic Detection of Texture Defects using Texture-Periodicity and Gabor Wavelets

Automatic Detection of Texture Defects using Texture-Periodicity and Gabor Wavelets Abstract Automatic Detection of Texture Defects using Texture-Periodicity and Gabor Wavelets V Asha 1, N U Bhajantri, and P Nagabhushan 3 1 New Horizon College of Engineering, Bangalore, Karnataka, India

More information

An Acceleration Scheme to The Local Directional Pattern

An Acceleration Scheme to The Local Directional Pattern An Acceleration Scheme to The Local Directional Pattern Y.M. Ayami Durban University of Technology Department of Information Technology, Ritson Campus, Durban, South Africa ayamlearning@gmail.com A. Shabat

More information

I. INTRODUCTION. Image Acquisition. Denoising in Wavelet Domain. Enhancement. Binarization. Thinning. Feature Extraction. Matching

I. INTRODUCTION. Image Acquisition. Denoising in Wavelet Domain. Enhancement. Binarization. Thinning. Feature Extraction. Matching A Comparative Analysis on Fingerprint Binarization Techniques K Sasirekha Department of Computer Science Periyar University Salem, Tamilnadu Ksasirekha7@gmail.com K Thangavel Department of Computer Science

More information

An Adaptive Histogram Equalization Algorithm on the Image Gray Level Mapping *

An Adaptive Histogram Equalization Algorithm on the Image Gray Level Mapping * Available online at www.sciencedirect.com Physics Procedia 25 (2012 ) 601 608 2012 International Conference on Solid State Devices and Materials Science An Adaptive Histogram Equalization Algorithm on

More information

Object Extraction Using Image Segmentation and Adaptive Constraint Propagation

Object Extraction Using Image Segmentation and Adaptive Constraint Propagation Object Extraction Using Image Segmentation and Adaptive Constraint Propagation 1 Rajeshwary Patel, 2 Swarndeep Saket 1 Student, 2 Assistant Professor 1 2 Department of Computer Engineering, 1 2 L. J. Institutes

More information

Fingerprint Recognition using Texture Features

Fingerprint Recognition using Texture Features Fingerprint Recognition using Texture Features Manidipa Saha, Jyotismita Chaki, Ranjan Parekh,, School of Education Technology, Jadavpur University, Kolkata, India Abstract: This paper proposes an efficient

More information

IMPLEMENTATION OF SPATIAL FUZZY CLUSTERING IN DETECTING LIP ON COLOR IMAGES

IMPLEMENTATION OF SPATIAL FUZZY CLUSTERING IN DETECTING LIP ON COLOR IMAGES IMPLEMENTATION OF SPATIAL FUZZY CLUSTERING IN DETECTING LIP ON COLOR IMAGES Agus Zainal Arifin 1, Adhatus Sholichah 2, Anny Yuniarti 3, Dini Adni Navastara 4, Wijayanti Nurul Khotimah 5 1,2,3,4,5 Department

More information

Computers and Mathematics with Applications. An embedded system for real-time facial expression recognition based on the extension theory

Computers and Mathematics with Applications. An embedded system for real-time facial expression recognition based on the extension theory Computers and Mathematics with Applications 61 (2011) 2101 2106 Contents lists available at ScienceDirect Computers and Mathematics with Applications journal homepage: www.elsevier.com/locate/camwa An

More information

CORRELATION BASED CAR NUMBER PLATE EXTRACTION SYSTEM

CORRELATION BASED CAR NUMBER PLATE EXTRACTION SYSTEM CORRELATION BASED CAR NUMBER PLATE EXTRACTION SYSTEM 1 PHYO THET KHIN, 2 LAI LAI WIN KYI 1,2 Department of Information Technology, Mandalay Technological University The Republic of the Union of Myanmar

More information

A New Feature Local Binary Patterns (FLBP) Method

A New Feature Local Binary Patterns (FLBP) Method A New Feature Local Binary Patterns (FLBP) Method Jiayu Gu and Chengjun Liu The Department of Computer Science, New Jersey Institute of Technology, Newark, NJ 07102, USA Abstract - This paper presents

More information

An algorithmic method to extend TOPSIS for decision-making problems with interval data

An algorithmic method to extend TOPSIS for decision-making problems with interval data Applied Mathematics and Computation 175 (2006) 1375 1384 www.elsevier.com/locate/amc An algorithmic method to extend TOPSIS for decision-making problems with interval data G.R. Jahanshahloo, F. Hosseinzadeh

More information

A new predictive image compression scheme using histogram analysis and pattern matching

A new predictive image compression scheme using histogram analysis and pattern matching University of Wollongong Research Online University of Wollongong in Dubai - Papers University of Wollongong in Dubai 00 A new predictive image compression scheme using histogram analysis and pattern matching

More information

Cellular Learning Automata-Based Color Image Segmentation using Adaptive Chains

Cellular Learning Automata-Based Color Image Segmentation using Adaptive Chains Cellular Learning Automata-Based Color Image Segmentation using Adaptive Chains Ahmad Ali Abin, Mehran Fotouhi, Shohreh Kasaei, Senior Member, IEEE Sharif University of Technology, Tehran, Iran abin@ce.sharif.edu,

More information

IDENTIFYING GEOMETRICAL OBJECTS USING IMAGE ANALYSIS

IDENTIFYING GEOMETRICAL OBJECTS USING IMAGE ANALYSIS IDENTIFYING GEOMETRICAL OBJECTS USING IMAGE ANALYSIS Fathi M. O. Hamed and Salma F. Elkofhaifee Department of Statistics Faculty of Science University of Benghazi Benghazi Libya felramly@gmail.com and

More information

A Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation

A Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation , pp.162-167 http://dx.doi.org/10.14257/astl.2016.138.33 A Novel Image Super-resolution Reconstruction Algorithm based on Modified Sparse Representation Liqiang Hu, Chaofeng He Shijiazhuang Tiedao University,

More information

Fuzzy Segmentation. Chapter Introduction. 4.2 Unsupervised Clustering.

Fuzzy Segmentation. Chapter Introduction. 4.2 Unsupervised Clustering. Chapter 4 Fuzzy Segmentation 4. Introduction. The segmentation of objects whose color-composition is not common represents a difficult task, due to the illumination and the appropriate threshold selection

More information

Embedded Descendent-Only Zerotree Wavelet Coding for Image Compression

Embedded Descendent-Only Zerotree Wavelet Coding for Image Compression Embedded Descendent-Only Zerotree Wavelet Coding for Image Compression Wai Chong Chia, Li-Minn Ang, and Kah Phooi Seng Abstract The Embedded Zerotree Wavelet (EZW) coder which can be considered as a degree-0

More information

CHAPTER-1 INTRODUCTION

CHAPTER-1 INTRODUCTION CHAPTER-1 INTRODUCTION 1.1 Fuzzy concept, digital image processing and application in medicine With the advancement of digital computers, it has become easy to store large amount of data and carry out

More information

Learning based face hallucination techniques: A survey

Learning based face hallucination techniques: A survey Vol. 3 (2014-15) pp. 37-45. : A survey Premitha Premnath K Department of Computer Science & Engineering Vidya Academy of Science & Technology Thrissur - 680501, Kerala, India (email: premithakpnath@gmail.com)

More information

CAD SYSTEM FOR AUTOMATIC DETECTION OF BRAIN TUMOR THROUGH MRI BRAIN TUMOR DETECTION USING HPACO CHAPTER V BRAIN TUMOR DETECTION USING HPACO

CAD SYSTEM FOR AUTOMATIC DETECTION OF BRAIN TUMOR THROUGH MRI BRAIN TUMOR DETECTION USING HPACO CHAPTER V BRAIN TUMOR DETECTION USING HPACO CHAPTER V BRAIN TUMOR DETECTION USING HPACO 145 CHAPTER 5 DETECTION OF BRAIN TUMOR REGION USING HYBRID PARALLEL ANT COLONY OPTIMIZATION (HPACO) WITH FCM (FUZZY C MEANS) 5.1 PREFACE The Segmentation of

More information

ISSN: X Impact factor: 4.295

ISSN: X Impact factor: 4.295 ISSN: 2454-132X Impact factor: 4.295 (Volume3, Issue1) Available online at: www.ijariit.com Performance Analysis of Image Clustering Algorithm Applied to Brain MRI Kalyani R.Mandlik 1, Dr. Suresh S. Salankar

More information

Change Detection in Remotely Sensed Images Based on Image Fusion and Fuzzy Clustering

Change Detection in Remotely Sensed Images Based on Image Fusion and Fuzzy Clustering International Journal of Electronics Engineering Research. ISSN 0975-6450 Volume 9, Number 1 (2017) pp. 141-150 Research India Publications http://www.ripublication.com Change Detection in Remotely Sensed

More information

Haresh D. Chande #, Zankhana H. Shah *

Haresh D. Chande #, Zankhana H. Shah * Illumination Invariant Face Recognition System Haresh D. Chande #, Zankhana H. Shah * # Computer Engineering Department, Birla Vishvakarma Mahavidyalaya, Gujarat Technological University, India * Information

More information

PROBLEM FORMULATION AND RESEARCH METHODOLOGY

PROBLEM FORMULATION AND RESEARCH METHODOLOGY PROBLEM FORMULATION AND RESEARCH METHODOLOGY ON THE SOFT COMPUTING BASED APPROACHES FOR OBJECT DETECTION AND TRACKING IN VIDEOS CHAPTER 3 PROBLEM FORMULATION AND RESEARCH METHODOLOGY The foregoing chapter

More information

Automatic Texture Segmentation for Texture-based Image Retrieval

Automatic Texture Segmentation for Texture-based Image Retrieval Automatic Texture Segmentation for Texture-based Image Retrieval Ying Liu, Xiaofang Zhou School of ITEE, The University of Queensland, Queensland, 4072, Australia liuy@itee.uq.edu.au, zxf@itee.uq.edu.au

More information

Temperature Calculation of Pellet Rotary Kiln Based on Texture

Temperature Calculation of Pellet Rotary Kiln Based on Texture Intelligent Control and Automation, 2017, 8, 67-74 http://www.scirp.org/journal/ica ISSN Online: 2153-0661 ISSN Print: 2153-0653 Temperature Calculation of Pellet Rotary Kiln Based on Texture Chunli Lin,

More information

Redefining and Enhancing K-means Algorithm

Redefining and Enhancing K-means Algorithm Redefining and Enhancing K-means Algorithm Nimrat Kaur Sidhu 1, Rajneet kaur 2 Research Scholar, Department of Computer Science Engineering, SGGSWU, Fatehgarh Sahib, Punjab, India 1 Assistant Professor,

More information

Multidirectional 2DPCA Based Face Recognition System

Multidirectional 2DPCA Based Face Recognition System Multidirectional 2DPCA Based Face Recognition System Shilpi Soni 1, Raj Kumar Sahu 2 1 M.E. Scholar, Department of E&Tc Engg, CSIT, Durg 2 Associate Professor, Department of E&Tc Engg, CSIT, Durg Email:

More information