Image Frame Fusion using 3D Anisotropic Diffusion

Size: px
Start display at page:

Download "Image Frame Fusion using 3D Anisotropic Diffusion"

Transcription

1 Image Frame Fusion using 3D Anisotropic Diffusion Fatih Kahraman 1, C. Deniz Mendi 1, Muhittin Gökmen 2 1 TUBITAK Marmara Research Center, Informatics Institute, Kocaeli, Turkey 2 ITU Computer Engineering Department, 34469, Istanbul, Turkey {fatih.kahraman, deniz.mendi}@bte.mam.gov.tr, gokmen@itu.edu.tr Abstract In this paper, a modified 3D anisotropic diffusion method is proposed to improve the multi-frame image fusion performance. Multi frame image sequence is considered to be composed of aligned and warped images. The goal of this approach is to obtain a restored image from the aligned and warped image sequence, where alignment error and Gaussian noise are reduced. The proposed method consists of medium band stack filter and tree-structured 3D diffusion filter. I. INTRODUCTION The usage of the surveillance camera is rapidly increasing day by day. For any security reason, it may be required to obtain one single and improved image either in resolution or in visual quality from different scenes of the video data. In addition, a poor quality video data suffering from investigation of a certain object may be utilized to get several images from the same scene and after processing for the enhancement of these images, it is possible to get a better quality picture. Several recent studies are centred on this issue [1-8]. The most successful methods for the spatiotemporal image restoration are increasingly used to enhance the quality of low-resolution video or image sequences. Such applications are crucial and bring an important add-on to forensic investigation as evidence [3]. In commercial applications, such as face recognition, car plate identification or any object investigation, a common approach is to use the spatio-temporal information of the video sequence. In order to increase the performance of such applications, it is possible to derive the restored image information from a multi frame image sequence which may be gathered from the consecutive frames of different scenes. In this work, we focus on restoring the image from a multi frame image sequence. The multi frame is supposed to be composed of the frames which are gathered from consecutive frames of different scenes of the video data. The objects in these frames are all aligned and warped to same reference shape. Therefore, it is expected to have alignment artefacts and additive noise which will be removed by 3D anisotropic diffusion filtering. In section 2, the warping algorithm, frame fusion and 3D diffusion process are discussed. Section 3 is devoted to the proposed method. The results of the proposed method are presented in Section 4. Section 5 describes the future plans and conclusions. II. MULTI-FRAME IMAGE RESTORATION The multi frame image restoration can be achieved in different ways. One is a data-driven approach where several image frames of the same objects are aligned to the reference frame. Aligned image sequence restoration can be established after eliminating variation stemming from pose and image perspective. This elimination is in the current work carried out by Annotating prominent object (i.e. face) features, (See Figure 1.a). Filtering out effects stemming from affine variations (translation, rotation and scaling), by a piece-wise affine warp onto a reference shape. Following this step, restoring the image from a multi frame image sequence can now be achieved easily by using of these shape-compensated images. Notice that multi frame is supposed to be composed of the frames which are gathered from consecutive frames of different scenes and they are warped to the same reference shape. All warped image frames are composed to obtain an image volume (See Figure 1.c). A. Image Warping The object under investigation in a video stream is modelled by a triangular based approach. The shape of the object is labelled by the fudicial points (landmarks). Relying upon the landmarks, a triangulated mesh is produced for the reference position and orientation of the object. By means of the image sequence, a piecewise affine warping [8][9] [10] is defined between corresponding triangles (See Figure 1.a and 1.b). In this study, warping is a crucial step, since image volume cannot be created unless the individual frames are warped. In particular, a piecewise affine warping is defined for a pair of triangles. Considering that for a triangle in the training mesh set of, there exists a corresponding triangle in mesh. This can be summarized; For any pixel, in, determine the triangle in which it is suited. Warp, to the triangle in, by means of the affine warping. Affine warping rules are defined by an input image sequence. In real implementation, it is possible to insert fudicial points shifts due to the poor resolution or high noise in the image. This causes alignment errors to the warped images.

2 B. Frame Averaging The multi frame image enhancement can be achieved by averaging the frames of the image sequence. This is a useful technique for reducing the additive noise. However, frame fusion techniques which are especially based on averaging methods cause smoothing effect. Therefore, significant deblurring improvement can further be achieved by image restoration filters. The multi frame averaging can thus be expressed as follows;, 1,, 1 where denotes each frame of the multi frame sequence. The number of frames in multi frame sequence is taken as. is the spatial deblurring filtering applied to each. The averaged frame is denoted by. Here, is taken as median filtering, Wiener filtering, 2D anisotropic diffusion filtering and unweighted averaging where is taken as identity function. The results of these methods are discussed in the Section 4, Experimental Results. Figure 1: Generation of image volume using warped image frame; a) annotated landmark on image frames, b) triangulated mesh based on the landmarks, c) image volume is formed up using warped images. (The right most figure in (c) shows half of the image volume) The frame averaging has the benefit of reducing the additive noise whereas Wiener filter deblurs the image by means of the point spread function which represents the blurring characteristics. Similarly, the other non-linear denoising filters, such as median and anisotropic diffusion filters, are also implemented in order to enhance the high frequencies of the image. On the other hand, removal of the additive noise by averaging process results in smoothing the high spatial resolution. Regarding that each frame is corrupted by zero-mean additive Gaussian noise with variance, the estimate of variance equals to, where stands for the number of frames. Decreasing the variance of the restored image as a result of the averaging brings smoothing effect whilst details are (a) (b) (c) often enhanced by deblurring filters. This two-fold problem gets more complicated for aligned and warped multi frames due to the incorrectly located fudicial marks and geometric deformations caused by warping methods. The later sections are devoted to handle such typical problems. C. 3D Anisotropic Diffusion Filter Anisotropic diffusion filter gives the flexibility of smoothing the image while keeping the high frequencies. Relying upon this property, anisotropic diffusion filter is implemented for restoration of the multi frame misaligned image sequence with additive noise. The diffusion process equilibrates the concentration differences throughout the image or similarly volume surface. The nonlinear 3D anisotropic diffusion filter equation is expressed below;,,,,,, (2),,,,,,,,, (3) where,,, is multi frame image sequence whereas is for the iteration time, not for the temporal indices. In the physical process of diffusion, corresponds to the flow function. is controlled by conduction coefficient and gradient vector. The flow rate in diffusion equations is controlled by the conduction coefficient by means of the flow constant, such that flow increases with the gradient strength,. This property of diffusion process is crucial in image enhancement. The flow is increased in smooth regions, where. On the other hand, the edges of the image are preserved by limiting the flow rate where. Therefore, behaves as the level under which the image is smoothed and over which the edges are preserved. In the meantime, noise is filtered based on the local structures of the image. This yields increasing the signal-tonoise ratio with no significant distortions through the edges. The anisotropic diffusion filtering has also some drawbacks. The diffusion process is somewhat an averaging process over the neighboring pixels. Due to this behavior, the image gets smoother as the iteration step increases or similarly the diffusion process denoted by in Eq(2) and (3). In this paper, 3D anisotropic nonlinear diffusion is implemented to multi frame image sequence. Each frame is warped from original images and they are aligned in a multi frame sequence. It is not considered that a temporal motion occurs between the frames. On contrary, it is considered that the edges are shifted due to the warping errors in which it is expected most likely to have jitter-type error around the edges of the image. Taking into consideration both the warping error and the additive Gaussian noise in multi frame sequence, 3D anisotropic diffusion filter is proposed to derive one single restored image. Since the 3D diffusion filter decreases the smoothing operation at the boundaries of the object, this can be exploited by setting different values to each unit of the Cartesian coordinates.

3 III. TREE-STRUCTURED 3D DIFFUSION FILTERING In this paper, the aligned image sequence is concerned for image restoration. Each frame of the aligned image sequence has an object which is warped to a certain position by means of indicated fudicial points on the original image. Therefore, incorrect locating of fudicial point results in wrong warping and the geometry will somehow be deformed. This will also yield that the edges of the warped images through the coordinate of,, consists of jitter error. This is obvious due to the fact that the fudicial marks are mostly located along the boundaries of the objects. A. Medium Band Stack Filter In order to improve the performance of the 3D diffusion filter, the image pixels are sorted along the -coordinate of the multi frame image sequence and the far end frames of the pixel-based re-ordered image sequence is truncated. This approach is named as Medium Band Stack Filter (MSB) by the authors of the paper. MBS is expressed as follows;,,,,, 1,,; 1,,; 1,, (4),,,,,,, 1 (5) operation indicates the sorting in an ascending order. By means of the sorting process, the one-dimensional vector along the -coordinate is sorted. In the following step as expressed in Eq(5), two certain slices from the each end of the image sequence which are denoted by, are removed out. The image sequence size is therefore truncated to by by 1 2 volume vector. One of the advantage of the medium band stack filtering as a pre-processing step before 3D diffusion filtering is the removal of the impulsive errors from the image sequence. Due to the sorting in an ascending order, the cropped away frames are almost the dark and light frames with insignificant object textures. Therefore relatively less critical information and highly noisy frames are removed. Another advantage maintained by sorting operator is that include the nearest intensity along the -coordinate due to the sorting operation and thus 3D diffusion process adapts itself to the steeping intensity levels not for the intensity variations caused by alignment and warping errors. B. Tree-Structured Diffusion Filtering The three-dimensional diffusion process is implemented to the cropped multi frame image volume,. It is empirically observed that partitioning the volume into sub regions along the -axis and carrying out the diffusion process at separate regions improves enhancement quality in terms of signal-to-noise ratio. The diffusion filter representation is denoted as follows;,,,, (6) signifies the i th diffusion process implemented to the region R j, where L j is the number of frames in R j. The 3D diffusion filter kernel is chosen as 3x3x3 volume. The conduction coefficient in Eq(3) is defined for 26 directions, separately. The dependence of is developed such that for the three main axes, there exists and. The coefficients are controlled by these two values. Based on the notation given in Eq(6), the tree-structured diffusion process is depicted in terms of the following steps; 1. The truncated image sequence volume is divided into three sub-regions. The boundaries of the regions are due to the locations where the second derivative of the image covariance is vanished. Each region is diffused;,,,,, (7) are derived. Throughout this step of algorithm, is chosen a high value in order to smooth out along the -axis. 2. The second step is focused on smoothing the side regions with respect to each other and thus a new reordering of the side regions is proposed as it follows;,,,,,,,,,,,, (8) and are the number of frames in the regions and, respectively. The second step diffusion process attempts to conduct flow along the far pixel values at equal distance to mid-point. Therefore they attempt to equilibrate each other. Following this step of diffusion, the volume frames are reordered to their original positions. This step of diffusion process is depicted as, and, 3. In the last step of the algorithm, all processed regions are diffused and this step is represented by the following expression;,,,,, (9) This step aims to enhance the image sequence by merging all diffusion processes of all regions. Tree-structured anisotropic diffusion filtering output is averaged in order to provide single restored image. One benefit of the tree structured algorithm is not only to get one solution but also the sub-region diffusion filter averaged images are possible to be alternative solutions in the context of image restoration. Therefore, it is possible to choose one of the best restored image among the listed ones;,,,,,, (10) where denotes the average operation as expressed in Eq(1). One of the drawbacks of this proposed method is that it requires at least 10 frames to be processed. It is however a useful technique for the aligned images from a video sequence where it is potentially available to obtain more samples of the object. In case of limited number of frames, it is possible to decrease the number of regions. On contrary, if there are enough amount of frames, it is possible to extend the tree branches by increasing the number of the regions.

4 (a) (b) (c) (d) (e) (f) (g) (h) Figure 2: The checker board test sequence. a) and b) are two successive frames representing the alignment error. The crossings of the white and black squares are randomly shifted within ±2 pixels. c) presents a randomly selected frame from the test sequence which is corrupted by Gaussian noise with variance of d) Averaging of 100 frames (SNR 5.56dB). e) Median filtering (SNR 7.56dB) and f) Wiener filter (SNR, 5.44dB). g) the average of the 2D diffusion filtering of frames of the test sequence (SNR 5.56dB). h) 3D diffusion filtering of the whole test sequence (SNR 5.51dB). (a) Figure 3: Tree-structured 3D Anisotropic Diffusion Filtering results are presented; a) presents output derived from all regions (10.08dB) b) is from Region 2 (11.43dB). IV. EXPERIMENTAL RESULTS In this section, we compare the distortion values obtained from the experimentation using three set of test data. The denoising performance of the proposed method is presented and discussed throughout this section while other image restoration techniques are also presented in order to give a comparison about the de-noising performance. In this context, frame averaging, median filtering, Wiener filtering, 2D/3D anisotropic diffusion filtering are implemented to the test data. A. Test Data a) Checker Board Data: Incorrect alignment is simulated by randomly shifting the boundaries of the squares up to 2 pixels. Test data is corrupted by Gaussian noise with variance of The number of the frames is 100 and the size of each frame is 160x160 pixels. Figure 2.a and 2.b represent the misalignment error. Figure 2.c shows a sample frame of the checker board within test video stream. b) Pointer Data: The scope of this test data is to present the performance improvement of the proposed method on a real video. The recorded video stream consists of two triangle forms and a line of text from several view angles (Fig 4.a). The video frames are corrupted by Gaussian noise with variance of 0.2. By means of the landmarks as shown in Figure 4.b, the alignment and warping the object under scope is achieved. The incorrect alignment due to the annotation error is depicted in Figure 4.c/d/e, which is likely undulation and unsharp edge effect. The test stream consists of 90 frames with the size of 150x140 pixels. c) Face Data: The purpose of using such a test data is to implement the proposed method to a real noisy video which has a more complex object. The person in recorded video (b) slowly turns his face by 45 o. Gaussian noise with 0.15 variance is contaminated to test stream. 73 landmarks as shown in Figure 1 is utilized in order to annotate the face. 38 frames with the size of 78x68 pixels are included within the test stream (Fig. 5.b) B. Performance Measurements The performance of the proposed method is measured by the signal-to-noise ratio. The SNR is defined as follows 10 log (11) where x is the observed image which is under performance evaluation and x gt is the ground-truth image. The groundtruth checker board data has no alignment error and no additive noise. On the other hand, the ground-truth image for the face data set and the pointer data set are regarded as the image on which each individual image from the test sequence is warped. Figure 4.a and 5.a depict the groundtruth image for the pointer and face test data, respectively. C. Results of Tree-Structured 3D Diffusion Filtering Before diffusion filtering, MBS (medium band stack) filtering is achieved on the board test data (See Figure 2). Following the MBS step, the checkerboard output image volume becomes such that the initial frame of the image sequence becomes a very dark image with larger black squares and through the -axis, white squares get larger and the pixel intensities increase to white level. The noisy parts of the image frames are pushed to the far sides of the frame sequence and therefore the less meaningful portions of the sorted image sequence are cropped away as indicated by in Eq(5). The choice of in all test data is 0.2. In this paper we proposed 3D diffusion method in a tree structured manner such that diffusion filtering is applied at each region or in other words at each level of the tree. Diffusion filtering is processed with separate set of and flow constant vector. The result of the proposed 3D diffusion filtering does not need to be necessarily a unique solution, but it is possible to choose the highest quality image among the sub-region diffusion filter outputs as well. This requires the experience of the user which is almost the case in the field of forensic evidence evaluation or medical imaging applications.

5 (a) (b) (c) (d) (e) (f) (g) (h) (i) Figure 4: a) is the ground truth image b) represents the landmarks utilized for triangulation c-d-e) are three samples of aligned images f) the input test sequence by which the performance measurements are handle. The variance of noise is 0.2 g) the conventional 3D diffusion filtering (6.33 db) h) the proposed filtering method result (7.26 db) i) Region 2 result (7.53 db). Image restoration performance of frame averaging (Fig 2.d), median filtering (Fig 2.e) and Wiener filtering (Fig 2.f) are presented in Figure 2. As in median filter approach, it is possible to show that dark and light regions of the image are recovered. However, the edges are smoothed along the misalignment region (Fig 2.e). Wiener and averaging performs the smoothing overall the image (Fig 2.d and 2.f). Figure 2.g and Figure 2.h represent that the averaging of 2D diffusion and the 3D diffusion perform quiet similar results. SNR values in Table 1, the frame averaging and Wiener filter performs better than median filtering with around 1dB of SNR improvement. (a) (b) (c) (d) Figure 5: The proposed method performance is presented by the face data. a) is the ground truth image, b) is the noise corrupted frame (var=0.15), c) Tree-structured 3-D Anisotropic Diffusion filtering result ( ), d) region 2 result for the tree-structured 3D diffusion filtering (, ). The nose and mount are zoomed and presented in the same order. The well-known image restoration methods perform a valuable amount improvement in SNR term. The SNRs of the checker board are distributed between 0.99 and 2.21 db. The pointer test data consists of image frames between 1.02 db and 3.12 db. In the context of face data, each frame of the face image sequence has an SNR measurement varying between 5.12 db and db. Similar improvements in SNR term is achieved by the image restoration methods which is approximately at least 3 db of SNR improvement. Table 1, tabulates the SNR observations for the checker board, pointer and face data set. The proposed method of 3D diffusion filtering is tested by three image sequences. Smoothing the intra-regions and preserving boundaries at the inter-regions property of diffusion filtering is emphasized by the proposed treestructured diffusion filtering. This can be observed on the checker board image that is shown in Figure 2. (a) Figure 6: a) The improvement in SNR between 0.05 and 0.7 noise variance and b) the length of frame sequence are compared. Triangle stands for the proposed tree structured 3D diffusion method, square for averaging and circle for conventional 3D diffusion method. Figure 3.a shows the overall output of tree-structured 3D diffusion, that is in Eq(9), whereas 3.b depicts,. Both of the figures have sharp edges and black/white region intensities are closer to their original levels. The SNR measurement for is 10.08dB and for, 11.43dB. This yields that the improvement in SNR measurement with respect to the frame averaging method is 4.52 db and 5.87 db, respectively. The pointer data set performs 7.26 db for, and 7.53 db for,, where it has to be remarked that conventional 3D anisotropic diffusion is about 6.30 db which is almost very close to the direct frame averaging method. Therefore, these SNR measurements designate that tree-structured 3D diffusion method introduces about 1.23 db of improvement above the conventional approach. The SNR improvement in pointer data set is less than the checker board data set. This issue is acceptable, since more severe alignment error is introduced. Fig 4.c, 4.d and 4.e are three aligned and warped image frames. Figure 4.g presents the conventional 3D diffusion filtering. Figure 4.h and 4.i are and,. The edges are more emphasized and the homogeneous regions are smoothed. The text line gets readable. The and, of face data set are shown in Figure 5.c and Figure 5.d, respectively. Both of the images are visually improved. The SNR measurements are also improved such as is db and db for,. The frame averaging SNR measurement is db. The visual inspection on the diffused images shows that the noise is (b)

6 filtered out while keeping the details on the both of the images. Based on the experimental results, it is observed that tree based 3D anisotropic diffusion combined with medium band stack filtering outperforms than the other methods for the multi frame image sequence which consists of alignment artifacts and significant amount of additive noise. Table 1. The SNR measurements obtained from proposed 3D diffusion filtering and from other image restoration methods. The SNR measurements are tabulated for checker board, pointer and face test data set. Methods Checker Board SNR Pointer Face Min individual frame SNR 0.99 db 1.02 db 5.12 db within the test stream Max individual frame SNR 2.21 db 3.12 db db within the test stream Frame Fusion (Average) 5.56 db 6.16 db db Median Filter 7.56 db 6.27 db db Wiener Filter 5.44 db 6,37 db db (x 0,σ 0.5 Wiener Filter 5.40 db 6.42 db db (x 0,σ D Anisotropic Diffusion 5.56 db 6.37 db db iteration 3-D Anisotropic Diffusion 5.51 db 6.33 db db iterations 3-D Anisotropic Diffusion 5.48 db 6.30 db db iterations Tree-structured 3D db 7.26 db db Anisotropic Diffusion Tree-structured 3D Anisotropic Diffusion (Region 2) db 7.53 db db The effect of noise variance to the performance of the tree-structured diffusion filtering is presented by Figure 6.a. As the noise variance gets above 0.1, approximately 2 db of additional SNR improvement can be observed with respect to the conventional methods such as frame averaging and 3D diffusion filtering which presumes no region based tree structure model. It is worth of noting that above 0.35 of noise variance, annotation process gets hard due to obscured landmarks. As the length of frame sequence increases, the SNR also improves. The frame length effect on performance improvement is shown in Figure 6.b. From 20 frames to 90 frames, approximately 1 db improvement can be observed. This outcome also presents that even down to 20 frames, the proposed method may have meaningful results. introduced to the sequence of images. The proposed method in future should be enriched for an adaptive way of region selection. Additionally, the image enhancement may be further improved relying upon the image content. However, the initial results show that this proposed method is a candidate for investigation of a certain object in case of gathering the information from different video scenes. ACKNOWLEDGMENT We are grateful to Dr. Binnur Kurt for his helpful comments and for his early contributions to our ideas. This work is supported by the National Scientific and Research Council of Turkey, project no: 108G002. REFERENCES [1] S. John, M. A. Vorontsov, Multi-frame Selective Information Fusion from Robust Error Estimation Theory, IEEE Trans. on Image Processing, Vol. 14 (5), pp , [2] F. Wheeler, X. Liu, and P. Tu, Multi-Frame Super-Resolution for Face Recognition, Proceeding of IEEE Conference on Biometrics: Theory, Applications and Systems (BTAS), pp , [3] D. Thomas, K. W. Bowyer, P. J. Flynn, Multi-frame Approaches To Improve Face Recognition, Proceedings of the IEEE Workshop on Motion and Video Computing (WMVC '07), pp , [4] M. K. Ozkan, A. T. Erdem, M. I. Sezan, A. M. Tekalp, Efficient multi-frame Wiener restoration of blurred and noisy image sequences, IP(1), No. 4, pp , [5] E. Dubois, S. Sabri, Noise reduction in image sequences using motion- compensated temporal filtering, IEEE Transactions on Communications, Vol. 32, pp , [6] B. K. Gunturk, Y. Altunbasak, R. M. Mersereau, Multi-frame information fusion for gray-scale and spatial enhancement of images, ICIP03, Vol. 2, pp , [7] INTEL, Video Image Reconstruction and Enhancement: A Terascale Computing Application, INTEL White Papers, [8] F. Kahraman, B. Kurt, M. Gokmen, Robust Face Alignment For Illumination and Pose Invariant Face Recognition, IEEE Conf. on Computer Vision and Pattern Recognition (CVPR 2007), Workshop on Biometrics, pp.1-7, [9] C. A. Glasbey and K. V. Mardia. A review of image warping methods. Journal of Applied Statistics, Vol. 25 (2), pp , [10] M. B. Stegmann, B. K. Ersbøll, R. Larsen. FAME - A Flexible Appearance Modeling Environment. IEEE Trans. Med. Imaging, Vol. 22(10), pp , V. CONCLUSION AND FUTURE WORKS We described a multi-frame image restoration method, based on the anisotropic diffusion process. The multi frame image sequence is supposed to be composed of warped and aligned images. We presented a tree-structured 3D diffusion filtering process which is combined with the medium band stack filtering. The proposed method improves the image restoration quality for the image sequence which is corrupted by additive noise and alignment artifacts are

Removing Atmospheric Turbulence

Removing Atmospheric Turbulence Removing Atmospheric Turbulence Xiang Zhu, Peyman Milanfar EE Department University of California, Santa Cruz SIAM Imaging Science, May 20 th, 2012 1 What is the Problem? time 2 Atmospheric Turbulence

More information

Face Hallucination Based on Eigentransformation Learning

Face Hallucination Based on Eigentransformation Learning Advanced Science and Technology etters, pp.32-37 http://dx.doi.org/10.14257/astl.2016. Face allucination Based on Eigentransformation earning Guohua Zou School of software, East China University of Technology,

More information

Structured Light II. Thanks to Ronen Gvili, Szymon Rusinkiewicz and Maks Ovsjanikov

Structured Light II. Thanks to Ronen Gvili, Szymon Rusinkiewicz and Maks Ovsjanikov Structured Light II Johannes Köhler Johannes.koehler@dfki.de Thanks to Ronen Gvili, Szymon Rusinkiewicz and Maks Ovsjanikov Introduction Previous lecture: Structured Light I Active Scanning Camera/emitter

More information

Chapter 3 Image Registration. Chapter 3 Image Registration

Chapter 3 Image Registration. Chapter 3 Image Registration Chapter 3 Image Registration Distributed Algorithms for Introduction (1) Definition: Image Registration Input: 2 images of the same scene but taken from different perspectives Goal: Identify transformation

More information

Illumination-Robust Face Recognition based on Gabor Feature Face Intrinsic Identity PCA Model

Illumination-Robust Face Recognition based on Gabor Feature Face Intrinsic Identity PCA Model Illumination-Robust Face Recognition based on Gabor Feature Face Intrinsic Identity PCA Model TAE IN SEOL*, SUN-TAE CHUNG*, SUNHO KI**, SEONGWON CHO**, YUN-KWANG HONG*** *School of Electronic Engineering

More information

Adaptive Wavelet Image Denoising Based on the Entropy of Homogenus Regions

Adaptive Wavelet Image Denoising Based on the Entropy of Homogenus Regions International Journal of Electrical and Electronic Science 206; 3(4): 9-25 http://www.aascit.org/journal/ijees ISSN: 2375-2998 Adaptive Wavelet Image Denoising Based on the Entropy of Homogenus Regions

More information

METRIC PLANE RECTIFICATION USING SYMMETRIC VANISHING POINTS

METRIC PLANE RECTIFICATION USING SYMMETRIC VANISHING POINTS METRIC PLANE RECTIFICATION USING SYMMETRIC VANISHING POINTS M. Lefler, H. Hel-Or Dept. of CS, University of Haifa, Israel Y. Hel-Or School of CS, IDC, Herzliya, Israel ABSTRACT Video analysis often requires

More information

Depth. Common Classification Tasks. Example: AlexNet. Another Example: Inception. Another Example: Inception. Depth

Depth. Common Classification Tasks. Example: AlexNet. Another Example: Inception. Another Example: Inception. Depth Common Classification Tasks Recognition of individual objects/faces Analyze object-specific features (e.g., key points) Train with images from different viewing angles Recognition of object classes Analyze

More information

x' = c 1 x + c 2 y + c 3 xy + c 4 y' = c 5 x + c 6 y + c 7 xy + c 8

x' = c 1 x + c 2 y + c 3 xy + c 4 y' = c 5 x + c 6 y + c 7 xy + c 8 1. Explain about gray level interpolation. The distortion correction equations yield non integer values for x' and y'. Because the distorted image g is digital, its pixel values are defined only at integer

More information

Segmentation and Tracking of Partial Planar Templates

Segmentation and Tracking of Partial Planar Templates Segmentation and Tracking of Partial Planar Templates Abdelsalam Masoud William Hoff Colorado School of Mines Colorado School of Mines Golden, CO 800 Golden, CO 800 amasoud@mines.edu whoff@mines.edu Abstract

More information

THE preceding chapters were all devoted to the analysis of images and signals which

THE preceding chapters were all devoted to the analysis of images and signals which Chapter 5 Segmentation of Color, Texture, and Orientation Images THE preceding chapters were all devoted to the analysis of images and signals which take values in IR. It is often necessary, however, to

More information

EXAM SOLUTIONS. Image Processing and Computer Vision Course 2D1421 Monday, 13 th of March 2006,

EXAM SOLUTIONS. Image Processing and Computer Vision Course 2D1421 Monday, 13 th of March 2006, School of Computer Science and Communication, KTH Danica Kragic EXAM SOLUTIONS Image Processing and Computer Vision Course 2D1421 Monday, 13 th of March 2006, 14.00 19.00 Grade table 0-25 U 26-35 3 36-45

More information

3D Computer Vision. Structured Light II. Prof. Didier Stricker. Kaiserlautern University.

3D Computer Vision. Structured Light II. Prof. Didier Stricker. Kaiserlautern University. 3D Computer Vision Structured Light II Prof. Didier Stricker Kaiserlautern University http://ags.cs.uni-kl.de/ DFKI Deutsches Forschungszentrum für Künstliche Intelligenz http://av.dfki.de 1 Introduction

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

BI-DIRECTIONAL AFFINE MOTION COMPENSATION USING A CONTENT-BASED, NON-CONNECTED, TRIANGULAR MESH

BI-DIRECTIONAL AFFINE MOTION COMPENSATION USING A CONTENT-BASED, NON-CONNECTED, TRIANGULAR MESH BI-DIRECTIONAL AFFINE MOTION COMPENSATION USING A CONTENT-BASED, NON-CONNECTED, TRIANGULAR MESH Marc Servais, Theo Vlachos and Thomas Davies University of Surrey, UK; and BBC Research and Development,

More information

Lecture 2 Image Processing and Filtering

Lecture 2 Image Processing and Filtering Lecture 2 Image Processing and Filtering UW CSE vision faculty What s on our plate today? Image formation Image sampling and quantization Image interpolation Domain transformations Affine image transformations

More information

Image Segmentation Techniques for Object-Based Coding

Image Segmentation Techniques for Object-Based Coding Image Techniques for Object-Based Coding Junaid Ahmed, Joseph Bosworth, and Scott T. Acton The Oklahoma Imaging Laboratory School of Electrical and Computer Engineering Oklahoma State University {ajunaid,bosworj,sacton}@okstate.edu

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

Image denoising using curvelet transform: an approach for edge preservation

Image denoising using curvelet transform: an approach for edge preservation Journal of Scientific & Industrial Research Vol. 3469, January 00, pp. 34-38 J SCI IN RES VOL 69 JANUARY 00 Image denoising using curvelet transform: an approach for edge preservation Anil A Patil * and

More information

SEMI-BLIND IMAGE RESTORATION USING A LOCAL NEURAL APPROACH

SEMI-BLIND IMAGE RESTORATION USING A LOCAL NEURAL APPROACH SEMI-BLIND IMAGE RESTORATION USING A LOCAL NEURAL APPROACH Ignazio Gallo, Elisabetta Binaghi and Mario Raspanti Universitá degli Studi dell Insubria Varese, Italy email: ignazio.gallo@uninsubria.it ABSTRACT

More information

Storage Efficient NL-Means Burst Denoising for Programmable Cameras

Storage Efficient NL-Means Burst Denoising for Programmable Cameras Storage Efficient NL-Means Burst Denoising for Programmable Cameras Brendan Duncan Stanford University brendand@stanford.edu Miroslav Kukla Stanford University mkukla@stanford.edu Abstract An effective

More information

Module 7 VIDEO CODING AND MOTION ESTIMATION

Module 7 VIDEO CODING AND MOTION ESTIMATION Module 7 VIDEO CODING AND MOTION ESTIMATION Lesson 20 Basic Building Blocks & Temporal Redundancy Instructional Objectives At the end of this lesson, the students should be able to: 1. Name at least five

More information

VIDEO DENOISING BASED ON ADAPTIVE TEMPORAL AVERAGING

VIDEO DENOISING BASED ON ADAPTIVE TEMPORAL AVERAGING Engineering Review Vol. 32, Issue 2, 64-69, 2012. 64 VIDEO DENOISING BASED ON ADAPTIVE TEMPORAL AVERAGING David BARTOVČAK Miroslav VRANKIĆ Abstract: This paper proposes a video denoising algorithm based

More information

arxiv: v1 [cs.cv] 28 Sep 2018

arxiv: v1 [cs.cv] 28 Sep 2018 Camera Pose Estimation from Sequence of Calibrated Images arxiv:1809.11066v1 [cs.cv] 28 Sep 2018 Jacek Komorowski 1 and Przemyslaw Rokita 2 1 Maria Curie-Sklodowska University, Institute of Computer Science,

More information

Image Restoration and Reconstruction

Image Restoration and Reconstruction Image Restoration and Reconstruction Image restoration Objective process to improve an image, as opposed to the subjective process of image enhancement Enhancement uses heuristics to improve the image

More information

Image Restoration and Reconstruction

Image Restoration and Reconstruction Image Restoration and Reconstruction Image restoration Objective process to improve an image Recover an image by using a priori knowledge of degradation phenomenon Exemplified by removal of blur by deblurring

More information

WEINER FILTER AND SUB-BLOCK DECOMPOSITION BASED IMAGE RESTORATION FOR MEDICAL APPLICATIONS

WEINER FILTER AND SUB-BLOCK DECOMPOSITION BASED IMAGE RESTORATION FOR MEDICAL APPLICATIONS WEINER FILTER AND SUB-BLOCK DECOMPOSITION BASED IMAGE RESTORATION FOR MEDICAL APPLICATIONS ARIFA SULTANA 1 & KANDARPA KUMAR SARMA 2 1,2 Department of Electronics and Communication Engineering, Gauhati

More information

Accurate 3D Face and Body Modeling from a Single Fixed Kinect

Accurate 3D Face and Body Modeling from a Single Fixed Kinect Accurate 3D Face and Body Modeling from a Single Fixed Kinect Ruizhe Wang*, Matthias Hernandez*, Jongmoo Choi, Gérard Medioni Computer Vision Lab, IRIS University of Southern California Abstract In this

More information

Disguised Face Identification (DFI) with Facial KeyPoints using Spatial Fusion Convolutional Network. Nathan Sun CIS601

Disguised Face Identification (DFI) with Facial KeyPoints using Spatial Fusion Convolutional Network. Nathan Sun CIS601 Disguised Face Identification (DFI) with Facial KeyPoints using Spatial Fusion Convolutional Network Nathan Sun CIS601 Introduction Face ID is complicated by alterations to an individual s appearance Beard,

More information

Comparative Analysis in Medical Imaging

Comparative Analysis in Medical Imaging 1 International Journal of Computer Applications (975 8887) Comparative Analysis in Medical Imaging Ashish Verma DCS, Punjabi University 1, Patiala, India Bharti Sharma DCS, Punjabi University 1, Patiala,

More information

AN ALGORITHM FOR BLIND RESTORATION OF BLURRED AND NOISY IMAGES

AN ALGORITHM FOR BLIND RESTORATION OF BLURRED AND NOISY IMAGES AN ALGORITHM FOR BLIND RESTORATION OF BLURRED AND NOISY IMAGES Nader Moayeri and Konstantinos Konstantinides Hewlett-Packard Laboratories 1501 Page Mill Road Palo Alto, CA 94304-1120 moayeri,konstant@hpl.hp.com

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

CHAPTER 3 IMAGE ENHANCEMENT IN THE SPATIAL DOMAIN

CHAPTER 3 IMAGE ENHANCEMENT IN THE SPATIAL DOMAIN CHAPTER 3 IMAGE ENHANCEMENT IN THE SPATIAL DOMAIN CHAPTER 3: IMAGE ENHANCEMENT IN THE SPATIAL DOMAIN Principal objective: to process an image so that the result is more suitable than the original image

More information

Linear Discriminant Analysis in Ottoman Alphabet Character Recognition

Linear Discriminant Analysis in Ottoman Alphabet Character Recognition Linear Discriminant Analysis in Ottoman Alphabet Character Recognition ZEYNEB KURT, H. IREM TURKMEN, M. ELIF KARSLIGIL Department of Computer Engineering, Yildiz Technical University, 34349 Besiktas /

More information

Statistical image models

Statistical image models Chapter 4 Statistical image models 4. Introduction 4.. Visual worlds Figure 4. shows images that belong to different visual worlds. The first world (fig. 4..a) is the world of white noise. It is the world

More information

Differential Structure in non-linear Image Embedding Functions

Differential Structure in non-linear Image Embedding Functions Differential Structure in non-linear Image Embedding Functions Robert Pless Department of Computer Science, Washington University in St. Louis pless@cse.wustl.edu Abstract Many natural image sets are samples

More information

CoE4TN3 Medical Image Processing

CoE4TN3 Medical Image Processing CoE4TN3 Medical Image Processing Image Restoration Noise Image sensor might produce noise because of environmental conditions or quality of sensing elements. Interference in the image transmission channel.

More information

Super-Resolution. Many slides from Miki Elad Technion Yosi Rubner RTC and more

Super-Resolution. Many slides from Miki Elad Technion Yosi Rubner RTC and more Super-Resolution Many slides from Mii Elad Technion Yosi Rubner RTC and more 1 Example - Video 53 images, ratio 1:4 2 Example Surveillance 40 images ratio 1:4 3 Example Enhance Mosaics 4 5 Super-Resolution

More information

Notes 9: Optical Flow

Notes 9: Optical Flow Course 049064: Variational Methods in Image Processing Notes 9: Optical Flow Guy Gilboa 1 Basic Model 1.1 Background Optical flow is a fundamental problem in computer vision. The general goal is to find

More information

Vivekananda. Collegee of Engineering & Technology. Question and Answers on 10CS762 /10IS762 UNIT- 5 : IMAGE ENHANCEMENT.

Vivekananda. Collegee of Engineering & Technology. Question and Answers on 10CS762 /10IS762 UNIT- 5 : IMAGE ENHANCEMENT. Vivekananda Collegee of Engineering & Technology Question and Answers on 10CS762 /10IS762 UNIT- 5 : IMAGE ENHANCEMENT Dept. Prepared by Harivinod N Assistant Professor, of Computer Science and Engineering,

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

REAL-TIME FACE SWAPPING IN VIDEO SEQUENCES: MAGIC MIRROR

REAL-TIME FACE SWAPPING IN VIDEO SEQUENCES: MAGIC MIRROR REAL-TIME FACE SWAPPING IN VIDEO SEQUENCES: MAGIC MIRROR Nuri Murat Arar1, Fatma Gu ney1, Nasuh Kaan Bekmezci1, Hua Gao2 and Hazım Kemal Ekenel1,2,3 1 Department of Computer Engineering, Bogazici University,

More information

Fast Noise Level Estimation from a Single Image Degraded with Gaussian Noise

Fast Noise Level Estimation from a Single Image Degraded with Gaussian Noise Fast Noise Level Estimation from a Single Image Degraded with Gaussian Noise Takashi Suzuki Keita Kobayashi Hiroyuki Tsuji and Tomoaki Kimura Department of Information and Computer Science, Kanagawa Institute

More information

Fast 3D Mean Shift Filter for CT Images

Fast 3D Mean Shift Filter for CT Images Fast 3D Mean Shift Filter for CT Images Gustavo Fernández Domínguez, Horst Bischof, and Reinhard Beichel Institute for Computer Graphics and Vision, Graz University of Technology Inffeldgasse 16/2, A-8010,

More information

Context based optimal shape coding

Context based optimal shape coding IEEE Signal Processing Society 1999 Workshop on Multimedia Signal Processing September 13-15, 1999, Copenhagen, Denmark Electronic Proceedings 1999 IEEE Context based optimal shape coding Gerry Melnikov,

More information

Spatio-Temporal Stereo Disparity Integration

Spatio-Temporal Stereo Disparity Integration Spatio-Temporal Stereo Disparity Integration Sandino Morales and Reinhard Klette The.enpeda.. Project, The University of Auckland Tamaki Innovation Campus, Auckland, New Zealand pmor085@aucklanduni.ac.nz

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

CS 231A Computer Vision (Fall 2012) Problem Set 3

CS 231A Computer Vision (Fall 2012) Problem Set 3 CS 231A Computer Vision (Fall 2012) Problem Set 3 Due: Nov. 13 th, 2012 (2:15pm) 1 Probabilistic Recursion for Tracking (20 points) In this problem you will derive a method for tracking a point of interest

More information

2D Image Morphing using Pixels based Color Transition Methods

2D Image Morphing using Pixels based Color Transition Methods 2D Image Morphing using Pixels based Color Transition Methods H.B. Kekre Senior Professor, Computer Engineering,MP STME, SVKM S NMIMS University, Mumbai,India Tanuja K. Sarode Asst.Professor, Thadomal

More information

SURVEY ON IMAGE PROCESSING IN THE FIELD OF DE-NOISING TECHNIQUES AND EDGE DETECTION TECHNIQUES ON RADIOGRAPHIC IMAGES

SURVEY ON IMAGE PROCESSING IN THE FIELD OF DE-NOISING TECHNIQUES AND EDGE DETECTION TECHNIQUES ON RADIOGRAPHIC IMAGES SURVEY ON IMAGE PROCESSING IN THE FIELD OF DE-NOISING TECHNIQUES AND EDGE DETECTION TECHNIQUES ON RADIOGRAPHIC IMAGES 1 B.THAMOTHARAN, 2 M.MENAKA, 3 SANDHYA VAIDYANATHAN, 3 SOWMYA RAVIKUMAR 1 Asst. Prof.,

More information

Probabilistic Facial Feature Extraction Using Joint Distribution of Location and Texture Information

Probabilistic Facial Feature Extraction Using Joint Distribution of Location and Texture Information Probabilistic Facial Feature Extraction Using Joint Distribution of Location and Texture Information Mustafa Berkay Yilmaz, Hakan Erdogan, Mustafa Unel Sabanci University, Faculty of Engineering and Natural

More information

Noise Reduction in Image Sequences using an Effective Fuzzy Algorithm

Noise Reduction in Image Sequences using an Effective Fuzzy Algorithm Noise Reduction in Image Sequences using an Effective Fuzzy Algorithm Mahmoud Saeid Khadijeh Saeid Mahmoud Khaleghi Abstract In this paper, we propose a novel spatiotemporal fuzzy based algorithm for noise

More information

Image Resizing Based on Gradient Vector Flow Analysis

Image Resizing Based on Gradient Vector Flow Analysis Image Resizing Based on Gradient Vector Flow Analysis Sebastiano Battiato battiato@dmi.unict.it Giovanni Puglisi puglisi@dmi.unict.it Giovanni Maria Farinella gfarinellao@dmi.unict.it Daniele Ravì rav@dmi.unict.it

More information

Motion Estimation for Video Coding Standards

Motion Estimation for Video Coding Standards Motion Estimation for Video Coding Standards Prof. Ja-Ling Wu Department of Computer Science and Information Engineering National Taiwan University Introduction of Motion Estimation The goal of video compression

More information

Enhanced Hemisphere Concept for Color Pixel Classification

Enhanced Hemisphere Concept for Color Pixel Classification 2016 International Conference on Multimedia Systems and Signal Processing Enhanced Hemisphere Concept for Color Pixel Classification Van Ng Graduate School of Information Sciences Tohoku University Sendai,

More information

ADAPTIVE DEBLURRING OF SURVEILLANCE VIDEO SEQUENCES THAT DETERIORATE OVER TIME. Konstantinos Vougioukas, Bastiaan J. Boom and Robert B.

ADAPTIVE DEBLURRING OF SURVEILLANCE VIDEO SEQUENCES THAT DETERIORATE OVER TIME. Konstantinos Vougioukas, Bastiaan J. Boom and Robert B. ADAPTIVE DELURRING OF SURVEILLANCE VIDEO SEQUENCES THAT DETERIORATE OVER TIME Konstantinos Vougioukas, astiaan J. oom and Robert. Fisher University of Edinburgh School of Informatics, 10 Crichton St, Edinburgh,

More information

CS 223B Computer Vision Problem Set 3

CS 223B Computer Vision Problem Set 3 CS 223B Computer Vision Problem Set 3 Due: Feb. 22 nd, 2011 1 Probabilistic Recursion for Tracking In this problem you will derive a method for tracking a point of interest through a sequence of images.

More information

Motion. 1 Introduction. 2 Optical Flow. Sohaib A Khan. 2.1 Brightness Constancy Equation

Motion. 1 Introduction. 2 Optical Flow. Sohaib A Khan. 2.1 Brightness Constancy Equation Motion Sohaib A Khan 1 Introduction So far, we have dealing with single images of a static scene taken by a fixed camera. Here we will deal with sequence of images taken at different time intervals. Motion

More information

Digital Image Processing. Prof. P. K. Biswas. Department of Electronic & Electrical Communication Engineering

Digital Image Processing. Prof. P. K. Biswas. Department of Electronic & Electrical Communication Engineering Digital Image Processing Prof. P. K. Biswas Department of Electronic & Electrical Communication Engineering Indian Institute of Technology, Kharagpur Lecture - 21 Image Enhancement Frequency Domain Processing

More information

Extensions of One-Dimensional Gray-level Nonlinear Image Processing Filters to Three-Dimensional Color Space

Extensions of One-Dimensional Gray-level Nonlinear Image Processing Filters to Three-Dimensional Color Space Extensions of One-Dimensional Gray-level Nonlinear Image Processing Filters to Three-Dimensional Color Space Orlando HERNANDEZ and Richard KNOWLES Department Electrical and Computer Engineering, The College

More information

Resolution Magnification Technique for Satellite Images Using DT- CWT and NLM

Resolution Magnification Technique for Satellite Images Using DT- CWT and NLM AUSTRALIAN JOURNAL OF BASIC AND APPLIED SCIENCES ISSN:1991-8178 EISSN: 2309-8414 Journal home page: www.ajbasweb.com Resolution Magnification Technique for Satellite Images Using DT- CWT and NLM 1 Saranya

More information

Texture Sensitive Image Inpainting after Object Morphing

Texture Sensitive Image Inpainting after Object Morphing Texture Sensitive Image Inpainting after Object Morphing Yin Chieh Liu and Yi-Leh Wu Department of Computer Science and Information Engineering National Taiwan University of Science and Technology, Taiwan

More information

Keywords: Thresholding, Morphological operations, Image filtering, Adaptive histogram equalization, Ceramic tile.

Keywords: Thresholding, Morphological operations, Image filtering, Adaptive histogram equalization, Ceramic tile. Volume 3, Issue 7, July 2013 ISSN: 2277 128X International Journal of Advanced Research in Computer Science and Software Engineering Research Paper Available online at: www.ijarcsse.com Blobs and Cracks

More information

Outline 7/2/201011/6/

Outline 7/2/201011/6/ Outline Pattern recognition in computer vision Background on the development of SIFT SIFT algorithm and some of its variations Computational considerations (SURF) Potential improvement Summary 01 2 Pattern

More information

Blur Space Iterative De-blurring

Blur Space Iterative De-blurring Blur Space Iterative De-blurring RADU CIPRIAN BILCU 1, MEJDI TRIMECHE 2, SAKARI ALENIUS 3, MARKKU VEHVILAINEN 4 1,2,3,4 Multimedia Technologies Laboratory, Nokia Research Center Visiokatu 1, FIN-33720,

More information

Real-time Detection of Illegally Parked Vehicles Using 1-D Transformation

Real-time Detection of Illegally Parked Vehicles Using 1-D Transformation Real-time Detection of Illegally Parked Vehicles Using 1-D Transformation Jong Taek Lee, M. S. Ryoo, Matthew Riley, and J. K. Aggarwal Computer & Vision Research Center Dept. of Electrical & Computer Engineering,

More information

Noise Model. Important Noise Probability Density Functions (Cont.) Important Noise Probability Density Functions

Noise Model. Important Noise Probability Density Functions (Cont.) Important Noise Probability Density Functions Others -- Noise Removal Techniques -- Edge Detection Techniques -- Geometric Operations -- Color Image Processing -- Color Spaces Xiaojun Qi Noise Model The principal sources of noise in digital images

More information

CS 231A Computer Vision (Winter 2014) Problem Set 3

CS 231A Computer Vision (Winter 2014) Problem Set 3 CS 231A Computer Vision (Winter 2014) Problem Set 3 Due: Feb. 18 th, 2015 (11:59pm) 1 Single Object Recognition Via SIFT (45 points) In his 2004 SIFT paper, David Lowe demonstrates impressive object recognition

More information

Outlines. Medical Image Processing Using Transforms. 4. Transform in image space

Outlines. Medical Image Processing Using Transforms. 4. Transform in image space Medical Image Processing Using Transforms Hongmei Zhu, Ph.D Department of Mathematics & Statistics York University hmzhu@yorku.ca Outlines Image Quality Gray value transforms Histogram processing Transforms

More information

Factorization with Missing and Noisy Data

Factorization with Missing and Noisy Data Factorization with Missing and Noisy Data Carme Julià, Angel Sappa, Felipe Lumbreras, Joan Serrat, and Antonio López Computer Vision Center and Computer Science Department, Universitat Autònoma de Barcelona,

More information

Head Frontal-View Identification Using Extended LLE

Head Frontal-View Identification Using Extended LLE Head Frontal-View Identification Using Extended LLE Chao Wang Center for Spoken Language Understanding, Oregon Health and Science University Abstract Automatic head frontal-view identification is challenging

More information

GRID WARPING IN TOTAL VARIATION IMAGE ENHANCEMENT METHODS. Andrey Nasonov, and Andrey Krylov

GRID WARPING IN TOTAL VARIATION IMAGE ENHANCEMENT METHODS. Andrey Nasonov, and Andrey Krylov GRID WARPING IN TOTAL VARIATION IMAGE ENHANCEMENT METHODS Andrey Nasonov, and Andrey Krylov Lomonosov Moscow State University, Moscow, Department of Computational Mathematics and Cybernetics, e-mail: nasonov@cs.msu.ru,

More information

Image Processing Lecture 10

Image Processing Lecture 10 Image Restoration Image restoration attempts to reconstruct or recover an image that has been degraded by a degradation phenomenon. Thus, restoration techniques are oriented toward modeling the degradation

More information

SYDE 575: Introduction to Image Processing

SYDE 575: Introduction to Image Processing SYDE 575: Introduction to Image Processing Image Enhancement and Restoration in Spatial Domain Chapter 3 Spatial Filtering Recall 2D discrete convolution g[m, n] = f [ m, n] h[ m, n] = f [i, j ] h[ m i,

More information

Image Coding with Active Appearance Models

Image Coding with Active Appearance Models Image Coding with Active Appearance Models Simon Baker, Iain Matthews, and Jeff Schneider CMU-RI-TR-03-13 The Robotics Institute Carnegie Mellon University Abstract Image coding is the task of representing

More information

Lecture 4: Spatial Domain Transformations

Lecture 4: Spatial Domain Transformations # Lecture 4: Spatial Domain Transformations Saad J Bedros sbedros@umn.edu Reminder 2 nd Quiz on the manipulator Part is this Fri, April 7 205, :5 AM to :0 PM Open Book, Open Notes, Focus on the material

More information

Region-based Segmentation

Region-based Segmentation Region-based Segmentation Image Segmentation Group similar components (such as, pixels in an image, image frames in a video) to obtain a compact representation. Applications: Finding tumors, veins, etc.

More information

Fundamentals of Stereo Vision Michael Bleyer LVA Stereo Vision

Fundamentals of Stereo Vision Michael Bleyer LVA Stereo Vision Fundamentals of Stereo Vision Michael Bleyer LVA Stereo Vision What Happened Last Time? Human 3D perception (3D cinema) Computational stereo Intuitive explanation of what is meant by disparity Stereo matching

More information

Critique: Efficient Iris Recognition by Characterizing Key Local Variations

Critique: Efficient Iris Recognition by Characterizing Key Local Variations Critique: Efficient Iris Recognition by Characterizing Key Local Variations Authors: L. Ma, T. Tan, Y. Wang, D. Zhang Published: IEEE Transactions on Image Processing, Vol. 13, No. 6 Critique By: Christopher

More information

Motion Tracking and Event Understanding in Video Sequences

Motion Tracking and Event Understanding in Video Sequences Motion Tracking and Event Understanding in Video Sequences Isaac Cohen Elaine Kang, Jinman Kang Institute for Robotics and Intelligent Systems University of Southern California Los Angeles, CA Objectives!

More information

An Approach for Reduction of Rain Streaks from a Single Image

An Approach for Reduction of Rain Streaks from a Single Image An Approach for Reduction of Rain Streaks from a Single Image Vijayakumar Majjagi 1, Netravati U M 2 1 4 th Semester, M. Tech, Digital Electronics, Department of Electronics and Communication G M Institute

More information

Locally Adaptive Regression Kernels with (many) Applications

Locally Adaptive Regression Kernels with (many) Applications Locally Adaptive Regression Kernels with (many) Applications Peyman Milanfar EE Department University of California, Santa Cruz Joint work with Hiro Takeda, Hae Jong Seo, Xiang Zhu Outline Introduction/Motivation

More information

Robust biometric image watermarking for fingerprint and face template protection

Robust biometric image watermarking for fingerprint and face template protection Robust biometric image watermarking for fingerprint and face template protection Mayank Vatsa 1, Richa Singh 1, Afzel Noore 1a),MaxM.Houck 2, and Keith Morris 2 1 West Virginia University, Morgantown,

More information

2D rendering takes a photo of the 2D scene with a virtual camera that selects an axis aligned rectangle from the scene. The photograph is placed into

2D rendering takes a photo of the 2D scene with a virtual camera that selects an axis aligned rectangle from the scene. The photograph is placed into 2D rendering takes a photo of the 2D scene with a virtual camera that selects an axis aligned rectangle from the scene. The photograph is placed into the viewport of the current application window. A pixel

More information

Artifacts and Textured Region Detection

Artifacts and Textured Region Detection Artifacts and Textured Region Detection 1 Vishal Bangard ECE 738 - Spring 2003 I. INTRODUCTION A lot of transformations, when applied to images, lead to the development of various artifacts in them. In

More information

Ensemble registration: Combining groupwise registration and segmentation

Ensemble registration: Combining groupwise registration and segmentation PURWANI, COOTES, TWINING: ENSEMBLE REGISTRATION 1 Ensemble registration: Combining groupwise registration and segmentation Sri Purwani 1,2 sri.purwani@postgrad.manchester.ac.uk Tim Cootes 1 t.cootes@manchester.ac.uk

More information

DATA EMBEDDING IN TEXT FOR A COPIER SYSTEM

DATA EMBEDDING IN TEXT FOR A COPIER SYSTEM DATA EMBEDDING IN TEXT FOR A COPIER SYSTEM Anoop K. Bhattacharjya and Hakan Ancin Epson Palo Alto Laboratory 3145 Porter Drive, Suite 104 Palo Alto, CA 94304 e-mail: {anoop, ancin}@erd.epson.com Abstract

More information

Image Sampling and Quantisation

Image Sampling and Quantisation Image Sampling and Quantisation Introduction to Signal and Image Processing Prof. Dr. Philippe Cattin MIAC, University of Basel 1 of 46 22.02.2016 09:17 Contents Contents 1 Motivation 2 Sampling Introduction

More information

Analysis of Image and Video Using Color, Texture and Shape Features for Object Identification

Analysis of Image and Video Using Color, Texture and Shape Features for Object Identification IOSR Journal of Computer Engineering (IOSR-JCE) e-issn: 2278-0661,p-ISSN: 2278-8727, Volume 16, Issue 6, Ver. VI (Nov Dec. 2014), PP 29-33 Analysis of Image and Video Using Color, Texture and Shape Features

More information

Filtering Images. Contents

Filtering Images. Contents Image Processing and Data Visualization with MATLAB Filtering Images Hansrudi Noser June 8-9, 010 UZH, Multimedia and Robotics Summer School Noise Smoothing Filters Sigmoid Filters Gradient Filters Contents

More information

International Journal of Advance Engineering and Research Development

International Journal of Advance Engineering and Research Development Scientific Journal of Impact Factor (SJIF): 4.72 International Journal of Advance Engineering and Research Development Volume 4, Issue 11, November -2017 e-issn (O): 2348-4470 p-issn (P): 2348-6406 Comparative

More information

Face Tracking. Synonyms. Definition. Main Body Text. Amit K. Roy-Chowdhury and Yilei Xu. Facial Motion Estimation

Face Tracking. Synonyms. Definition. Main Body Text. Amit K. Roy-Chowdhury and Yilei Xu. Facial Motion Estimation Face Tracking Amit K. Roy-Chowdhury and Yilei Xu Department of Electrical Engineering, University of California, Riverside, CA 92521, USA {amitrc,yxu}@ee.ucr.edu Synonyms Facial Motion Estimation Definition

More information

A New Technique of Extraction of Edge Detection Using Digital Image Processing

A New Technique of Extraction of Edge Detection Using Digital Image Processing International OPEN ACCESS Journal Of Modern Engineering Research (IJMER) A New Technique of Extraction of Edge Detection Using Digital Image Processing Balaji S.C.K 1 1, Asst Professor S.V.I.T Abstract:

More information

Image Sampling & Quantisation

Image Sampling & Quantisation Image Sampling & Quantisation Biomedical Image Analysis Prof. Dr. Philippe Cattin MIAC, University of Basel Contents 1 Motivation 2 Sampling Introduction and Motivation Sampling Example Quantisation Example

More information

Motivation. Intensity Levels

Motivation. Intensity Levels Motivation Image Intensity and Point Operations Dr. Edmund Lam Department of Electrical and Electronic Engineering The University of Hong ong A digital image is a matrix of numbers, each corresponding

More information

A Robust Wipe Detection Algorithm

A Robust Wipe Detection Algorithm A Robust Wipe Detection Algorithm C. W. Ngo, T. C. Pong & R. T. Chin Department of Computer Science The Hong Kong University of Science & Technology Clear Water Bay, Kowloon, Hong Kong Email: fcwngo, tcpong,

More information

CS4442/9542b Artificial Intelligence II prof. Olga Veksler

CS4442/9542b Artificial Intelligence II prof. Olga Veksler CS4442/9542b Artificial Intelligence II prof. Olga Veksler Lecture 8 Computer Vision Introduction, Filtering Some slides from: D. Jacobs, D. Lowe, S. Seitz, A.Efros, X. Li, R. Fergus, J. Hayes, S. Lazebnik,

More information

Textural Features for Image Database Retrieval

Textural Features for Image Database Retrieval Textural Features for Image Database Retrieval Selim Aksoy and Robert M. Haralick Intelligent Systems Laboratory Department of Electrical Engineering University of Washington Seattle, WA 98195-2500 {aksoy,haralick}@@isl.ee.washington.edu

More information

Reconstructing Images of Bar Codes for Construction Site Object Recognition 1

Reconstructing Images of Bar Codes for Construction Site Object Recognition 1 Reconstructing Images of Bar Codes for Construction Site Object Recognition 1 by David E. Gilsinn 2, Geraldine S. Cheok 3, Dianne P. O Leary 4 ABSTRACT: This paper discusses a general approach to reconstructing

More information

Locally Weighted Least Squares Regression for Image Denoising, Reconstruction and Up-sampling

Locally Weighted Least Squares Regression for Image Denoising, Reconstruction and Up-sampling Locally Weighted Least Squares Regression for Image Denoising, Reconstruction and Up-sampling Moritz Baecher May 15, 29 1 Introduction Edge-preserving smoothing and super-resolution are classic and important

More information