CHAPTER 8 COMPOUND CHARACTER RECOGNITION USING VARIOUS MODELS

Size: px
Start display at page:

Download "CHAPTER 8 COMPOUND CHARACTER RECOGNITION USING VARIOUS MODELS"

Transcription

1 CHAPTER 8 COMPOUND CHARACTER RECOGNITION USING VARIOUS MODELS

2 8.1 Introduction The recognition systems developed so far were for simple characters comprising of consonants and vowels. But there is one more category of characters called as compound characters in Marathi script. These characters are formed by joining two or more consonants. The recognition of compound characters is still more difficult due to their features. In this paper, we propose a system for unconstrained handwritten Marathi compound character recognition without separation of the characters in the compound character. Some pieces of work for handwritten Bangla compound characters are found [117]. No work for handwritten Marathi compound character recognition is found so far to the best of our knowledge. This chapter discusses the design of various models; simple as well as hybrid for the recognition of handwritten Marathi compound characters. Section 8.2 describes the features of compound characters. Section 8.3 presents the data collection, pre-processing and structural classification which is common to all the models. Section 8.4 presents the system using template matching using various kernels. Section 8.5 discusses the design of system using neural networks while Section 8.6 describes a hybrid model for the same. Section 8.7 discusses the concluding remarks. 8.2 Features of compound characters Marathi script has a complex system of compound characters in which two or more consonants are combined forming a new special symbol as shown in Figure 8.1. Figure 8.1 Subset of Compound characters with different joining strategies The compound characters as seen in Figure 8.1 are formed by combination of some specific consonants in order to give a meaningful combination. The compound character can have two or more characters joined together in various ways as shown in 8. Compound Character Recognition using Various Models 133

3 Figure 8.1. One way of forming compound character is by removing the vertical line of a character and then joining it to the other on its left hand side. This type of joining is more common. Another way of connection of characters in the compound character is by just joining the characters side by side or one above the other. In another way, one of the characters completely changes its form and then gets connected to the other to form a compound character. More than two consonants also join in various ways to form a compound character. As seen from Figure 8.1, the compound characters not only exhibit a variation in the shape of the character but also in the aspect ratio as per the joining strategy. One might get tempted to use the features like aspect ratio or number of end points, but the various joining strategies limits the use of these features to achieve acceptable recognition accuracy. All these challenges cannot be met by just a single feature extractor or a single classifier. Hence a multistage system is needed that can recognize the characters over a wide range of varying conditions. Compound characters in Marathi script occur more frequently in the script as compared to other languages derived from Devanagari. The occurrence of compound characters in Marathi is found to be about 11 to 12% whereas in other scripts of Devanagari and Bangla script, it is just 5 to 7% (17). This percentage is calculated by going through books, newspapers and other literature in Marathi script. For example, the word in Hindi language (derived from Devanagari) is in Marathi. The word in Hindi is in Marathi. Also the usage of English words in Marathi further adds to the compound characters in the Marathi script. For example, the word computer is a more commonly used word instead of its Marathi meaning. Now the word computer when written in Marathi is, which includes a compound character as seen here. The next section discusses the compound characters used for recognition in the proposed system along with the pre-processing and structural classification steps. 8.3 Data Collection, Pre-processing and Structural classification In the proposed system, we aim at recognizing handwritten Marathi compound characters. This is done by employing multiple feature extraction and classification stages. At first, the character is pre-classified into one of 24 classes based upon the 8. Compound Character Recognition using Various Models 134

4 structural features in the same way as discussed in Chapter 5. This two stage structural classification is followed by character resizing. The character is resized to a typical size of 16x16 or 32x32 and then further given to the various systems discussed in the next sections. The handwritten compound characters used for recognition are shown in Figure 8.2. It comprises of 40 compound characters and 34 split characters resulting into 74 characters in all. Figure 8.2 Characters used in the proposed system The database for handwritten compound characters is created by scanning the characters at 300 dpi using a flatbed scanner. The images are stored in bmp file format. Here we assume that the consonants in the compound characters are either touching or overlapping. However, sometimes they may not be touching or overlapping or may result into a gap or separate after binarization and pre-processing. These split components of a compound character result into separate entities after segmentation. In order to consider these separate entities of the compound characters, we also consider the split components of the compound character for recognition Figure 8.3 shows few examples of compound characters split into two entities during pre-processing. Figure 8.3 Example of split characters 8. Compound Character Recognition using Various Models 135

5 The binarization, pre-processing and the structural classification is carried out in the same manner as discussed in Chapter 5. At the end of the structural classification, there are 24 classes with compound characters in them based upon the structural features. These now form the database for generation of kernels for template matching or for training the neural network. The handwritten Marathi compound character dataset was collected from different individuals. More than 500 samples per character were collected, resulting in about more than character samples. Out of these, two third of the characters were used for database creation for storing as templates and for training the neural network and rest were used for testing. No standard database is available for handwritten Devanagari compound characters so far. Figure 8.4 show some sample characters in the database stored after pre-processing. Figure 8.4 Pre-processed compound characters The compound characters classified to the 24 classes after pre-processing are shown in Table 8.1. Each structural class shows different number of characters obtained as per the writing style of the writers. The classes contain compound characters as well as split characters. Table 8.1 Number of compound characters in the structural class Sr. no. Class No. of characters classified Characters classified 1. NBE/ NBE/ Compound Character Recognition using Various Models 136

6 3. NBE/ NBE/ NBNE/ NBNE/ NBNE/ NBNE/11 9. MBE/ MBE/ MBE/10 MBE/ Compound Character Recognition using Various Models 137

7 13. MBNE/ MBNE/ MBNE/10 MBNE/ EBE/ EBE/ EBE/ EBE/ EBNE/ EBNE/ Compound Character Recognition using Various Models 138

8 23. EBNE/ EBNE/11 19 The compound characters as discussed are more complex and incorporate a large amount of details in them. These details are the character shape, endpoints, junction points, curves and strokes. Various recognition techniques are developed step-wise for the recognition of these characters and the results are analyzed. The recognition of compound characters is carried out by various models like: 1. Template matching 2. Neural networks, and 3. Hybrid network The cropped binary characters in the structural classes are used as database for template matching as well as training the neural network. The various types of models discussed in the previous sections for handwritten compound character recognition are executed on an Intel Core 2 Duo CPU running on 2GHz with 2 GB RAM. The next sections discuss the recognition techniques mentioned above in detail. 8. Compound Character Recognition using Various Models 139

9 8.4 Compound character recognition using template matching In this system, the characters are recognized using template matching. As discussed before, the basic template matching technique performs cross correlation of the image f (m, n) with the template g (m, n). The result contains peaks at the location of the matches between the template and the underlying object. This matching is indicated by r which is the value between 0 and 1, where higher the value, more is the similarity. ( f f )( g g) m n r = (8.1) 2 2 ( f f ) ( g g) m n m n where, f and g are the mean of the original image and the template respectively. Matching is done only with the kernels in the class to which the character is classified after two stage structural classification. This reduces the number of comparisons and speeds up the recognition. The templates used for template matching are of various types. They are as follows: 1. Binary templates 2. Convolved binary templates 3. Wavelet approximation templates and 4. Modified wavelet approximation templates. The block diagram for the template matching procedure is shown in Figure 8.5. As shown in Figure 8.5, the compound character is pre-processed, classified structurally and then resized to generate a template as per the types mentioned above and then matched with the similar templates in the database. The results obtained by each template matching procedure with the respective templates are analyzed. The procedure for generation of these templates is explained next. 8. Compound Character Recognition using Various Models 140

10 Character input Pre-processing Structural Classification Character Resizing Binary templates Convolved binary templates Wavelet Approximation templates Modified Wavelet Approximation templates Template matching Template matching Template matching Template matching Recognized character Recognized character Recognized character Recognized character Figure 8.5 Template matching using different templates Generation of binary templates Here the templates used for matching are the binary character images which are resized to either 16x16 or 32x Generation of convolved templates In this method, instead of using the resized binary images as templates directly, they are convolved to themselves and stored as templates. Convolution is just like correlation, except that the filter is flipped before correlating. Convolution performs filtering operation and generates a multidimensional space invariant to rotation. This is useful in improving the recognition accuracy with template matching. Convolution between an image I x, y and a template T x, y is defined as 8. Compound Character Recognition using Various Models 141

11 f g = ( x, y) W f ( x', y' ) g( i x', j y' ) (8.2) where, x ' = x + i and y ' = y + j. The equation for convolution shows that the template matrix is flipped in both horizontal and vertical direction before multiplying the overlapped input data. Convolution of both 16x16 and 32x32 images is carried out for matching Generation of wavelet approximation templates Wavelet transform is another technique used for template generation. The wavelet transform exhibits the features like separability, scalability, translatability, orthogonality and multiresolution capability. Single level wavelet decomposition is used for templategeneration. Given separable 2D scaling and wavelet functions, the scaled and translated basis function is defined as: j / 2 j j ϕ ( x, y) = 2 ϕ(2 x m,2 y ), (8.3) j, m, n n i j / 2 i j j ϕ ( x, y) = 2 ψ (2 x m,2 y ), (8.4) j, m, n n where index i identifies the directional wavelets i.e. horizontal, vertical and diagonal details. The discrete wavelet transform of function f(x, y) of size MxN is then M 1 N 1 1 Wϕ ( j0, m, n) = f ( x, y) ϕ j,, (, ) 0 m n x y (8.5) MN x= 0 y= 0 M 1 N 1 i 1 i Wψ ( j, m, n) = f ( x, y) ψ j, m, n ( x, y) (8.6) MN x= 0 y= 0 where, ( j 0, m, n) coefficients define an approximation of f(x, y) at scale j 0 and W i ψ W ϕ ( j, m, n) add horizontal, vertical and diagonal details for scales j j 0. The discrete wavelet transform can be implemented using digital filters and down samplers. The high pass or detail component characterizes the image s high-frequency information with vertical orientation; the low-pass, approximation component contains its low-frequency, vertical information. The approximation coefficients obtained for every character after single level decomposition is stored as the templates in the database. 8. Compound Character Recognition using Various Models 142

12 The characters are then normalized to a fixed size of 16x16 after structural classification for kernel generation. A single level wavelet decomposition of the resized character image generates the approximation and the detail coefficients. The decomposition is done with respect to Daubechies wavelet. Here, j 0 =0 and M=N=16 which is equal to 2 j, hence j=0, 1, 2,, 7 and m, n=0, 1, 2,., 15. Thus 8x8 approximation coefficients generated after single level wavelet decomposition are used to match with the kernels stored in the database obtained in the similar manner. The wavelet decomposition reduces the image size by two, which further reduces the computations and increases the speed of recognition. The characters are also resized to 32x32 to generate 16x16 approximation features which are used as templates. The results are analyzed for both Generation of modified wavelet approximation templates The convolved templates obtained by convolving the binary templates with themselves gave improved results over simply binary templates. This was an inspiration to obtain the modified wavelet approximation templates. Here the templates generated in the previous step are used as wavelet kernels. The approximation coefficients obtained using single level wavelet decomposition is convolved with themselves for generation of modified wavelet template. Convolution between a 2D signal f(x, y) and a template g(x, y) is defined as f g = ( x, y) W f ( x', y' ) g( i x', j y' ) (8.7) where, x =x + i and y =y + j. The convolved output as indicated above is stored as the template in the database for matching. C = conv2 (A, B) computes the two-dimensional convolution of matrices A and B in MATLAB. The size of C in each dimension is equal to the sum of the corresponding dimensions of the input matrices, minus one. That is, if the size of A is [ma, na] and the size of B is [mb, nb], then the size of C is [ma+mb-1, na+nb-1].the 16x16 and 32x32 images are convolved with themselves to generate 31x31 and 63x63 kernels respectively, which are used for matching. In this method, we are trying to generate a wavelet kernel using 2D wavelet decomposition for each character. When a separate kernel is generated 8. Compound Character Recognition using Various Models 143

13 for each character, the feature space is different for each character and it increases the interclass and intra class separation between the characters Results for compound character recognition using template matching The compound character recognition is done using four different types of templates as discussed before. The sample compound character shown in Figure 8.6 (tya) is used for recognition. This character got classified to EBNE/11 class after structural classification. This class contains 19 different characters with 564 samples in all. The recognition using all the four template matching methods with 16x16 and 32x32 resize factors respectively, is shown in Figure 8.7 (a) through (h). The graphs show total number of characters in the EBNE/11 class on the x-axis and the matching value r on the y-axis. Figures 8.7 (a), 8.7 (b) and 8.7 (e) give the maximum similarity at index 89, which shows misclassification of the characters. The character is misclassified as vya which closely resembles the character under test. In rest of the figures, the maximum similarity value is found at index 74, which gives the correct classification result. Figure 8.6 Sample Compound character used for testing (a) (b) 8. Compound Character Recognition using Various Models 144

14 (c) (d) (e) (f) (g) (h) Figure 8.7 Compound character similarity matching using different templates (a) Binary template with resize factor 16x16, (b) Binary template with resize factor 32x32, (c) Convolved template with resize factor 16x16, (d) Convolved template with resize factor 32x32, (e) Wavelet approximation template with resize factor 16x16, (f) Wavelet approximation template with resize factor 32x32, (g) Modified wavelet approximation template with resize factor 16x16, (h) Modified wavelet approximation template with resize factor 32x32 8. Compound Character Recognition using Various Models 145

15 The character is misclassified by both 16x16 and 32x32 binary template as shown in Figure 8.7 (a) and (b). While it is classified correctly by both the convolved templates with resize factor 16x16 and 32x32 respectively as per Figure 8.7 (c) and (d). The character is further misclassified by wavelet approximation template with resize factor 16x16 but correctly classified by wavelet approximation template with resize factor 32x32 as shown in Figures 8.7 (e) and (f) respectively. Finally the character is correctly classified by the modified wavelet approximation templates with both 16x16 and 32x32 resize factors as shown in Figure 8.7 (g) and (h) respectively. Table 8.2 Template matching performance for compound character recognition Sr. Template no. generation technique 1. Binary template Resize factor Template size Recognition Time for EBNE/11 Overall Recognition rate (%) (sec) 16x16 16x x32 32x Convolved binary template 3. Wavelet approximation template 4. Modified wavelet approximation template 16x16 31x x32 63x x16 8x x32 16x x16 15x x32 31x The results in the Table 8.2 show that as the template size and the computations increase, the time required for recognition goes on increasing. The recognition time is also a function of number of characters and the samples per character in the class. The recognition rate increases with the increase in the resize factor and the number of samples per character. Increasing the resize factor did not show considerable 8. Compound Character Recognition using Various Models 146

16 improvement in the recognition rate but certainly requires more time to execute. Hence only 16x16 and 32x32 resize factors are considered. 8.5 Compound character recognition using neural network 8.8. The flow of the recognition scheme using neural network is depicted in Figure Handwritten Compound character Handwritten Compound character Pre-processing Pre-processing Structural Classification Structural Classification Character Resizing Character Resizing Modified Wavelet features Modified Wavelet features Neural Network Classifier Neural Network Training Recognized character Testing Phase Save weights and biases Training Phase Figure 8.8 Compound character recognition using neural network It consists of training phase and testing phase. At first, the characters are preclassified based upon their structural features. A two stage structural classification 8. Compound Character Recognition using Various Models 147

17 technique is implemented in both the phases. As seen in the template matching technique, the modified wavelet approximation features gave the highest recognition accuracy amongst other template generation technique. This gave an inspiration to use these features with neural network to improve the recognition rate. In the training phase, modified wavelet featured are obtained by convolving wavelet approximation features and used for training the neural network built as per the set parameters, so as to fix the weights and biases for each character. In the testing phase, similar features are extracted from the character. This is done again after pre-processing and structural classification of the character. The features are applied as inputs to the neural network. The output of the neural network yields the final recognition result. The neural network implemented is multilayer perceptron with the parameters as given in Table Feature extraction for compound character recognition using neural network The 16x16 resized binary images of compound character are decomposed using 2D single level wavelet decomposition for extraction of approximation features. This procedure is explained in Section The 8x8 approximation coefficients thus obtained are convolved with themselves to generate 64 modified wavelet approximation features. The instruction used for convolution operation is Cs = conv2(a,b,'same'), where, Cs is of the same size as A. Thus the 8x8 approximation features are convolved with themselves to obtain 8x8 = 64 modified wavelet approximation features. These features are further applied for training and testing the neural network Neural network design for compound character recognition using neural network Multilayer perceptron is used for compound character recognition. The parameters selected for the neural network are given in Table Compound Character Recognition using Various Models 148

18 Table 8.3 Neural network parameter settings Parameters Number of inputs Values 64 : Modified wavelet approximation features Number of hidden layers 1 Number of neurons in hidden layer Equal to the square root of the product of number of inputs and number of outputs Hidden layer activation function hyperbolic tangent sigmoid transfer function Number of neurons in output layer Number of characters in the structural class Output layer activation function Linear Goal Error function mse Maximum number of epoch 300 Training algorithm Levenberg-Marquardt algorithm Results for compound character recognition using neural network The neural network performance with modified wavelet features is given in Table 8.4. As indicated in the table, overall recognition accuracy for handwritten Marathi compound character is %. This shows improved recognition accuracy over template matching technique. Also the time required to recognize a character is approximately 0.05 sec irrespective of the number of characters in the structural class, i.e. it is not dependent upon the number of characters or their samples in the structural class. This gives an additional advantage over template matching technique. Table 8.4 Neural network performance for compound character recognition Sr. no. Feature extraction technique Recognition technique Recognition rate (%) Modified wavelet features Neural network Compound Character Recognition using Various Models 149

19 8.6 Proposed smart model for handwritten Compound character recognition using hybrid network The structure and the need for hybrid model are studied in Section 7.2 and Section 7.3 respectively. Such a model is more essential in the case of compound characters, as both the complexity and the number of characters are increased. But it is also found in the hybrid models in Chapter 7 that, the template matching approach in the structure required much more time for recognition due to matching with all the templates in the database. Hence, in the hybrid network proposed for compound character recognition, template matching classifier is replaced by a neural network to optimize, both the accuracy and the speed. Character input Pre-processing Structural Classification Character Resizing Pixel Density Features Euclidean Distance Features Modified Wavelet Features Neural Network B Neural Network Neural Network A Majority voting Recognized character C Figure 8.9 Proposed smart model for compound character recognition using hybrid model 8. Compound Character Recognition using Various Models 150

20 Figure 8.9 shows the proposed smart model for handwritten compound character recognition using hybrid model. As shown in Figure 8.9, at first, the character is preclassified into one of 24 classes based upon the structural features. This two stage structural classification is followed by character resizing. The character is resized to a fixed size of 16x16 or 70x50 depending upon the feature extraction technique. Three different features are extracted from the resized character and applied to three neural networks built for each structural class. In the testing phase, the weights and biases, fixed during training, are used to get the recognition result from all the three branches as indicated by A, B and C in Figure 8.9. The final recognition decision is based upon majority voting criterion. The concept of majority voting is studied in Chapter 7. The designs indicated by paths A and B in Figure 8.9 are implemented individually in Chapter 5 while the design indicated by path C in Figure 8.9 is implemented in Chapter 6. The hybrid network is thus designed by maximum voting criteria applied to all the three paths. The design of this hybrid network is further studied in detail in the following sub-sections Feature extraction for compound character recognition using hybrid network The resized binary compound characters are used to extract three different types of features as indicated in Table 8.5. Table 8.5 Feature extraction parameter settings for the hybrid network Sr. Feature extraction Resize No. of features no. technique factor 1. Normalized pixel density features 70x50 35 (character partitioning into 35 nonoverlapping blocks of size 100x100) 2. Euclidean distance features 16x16 32 (16 in horizontal and 16 in vertical directions) 3. Modified wavelet approximation features 16x16 64 (obtained after convolving 8x8 wavelet approximation features) The procedure for extraction of normalized pixel density features and Euclidean distance features is explained in Chapter 5, Section The procedure for 8. Compound Character Recognition using Various Models 151

21 extraction of modified wavelet approximation features is discussed in Section These features are applied to three different neural networks in the hybrid structure. The number of inputs for each network is differs as per the number of features extracted. The architecture of the neural networks is discussed next Neural network design for compound character recognition using hybrid network The hybrid network for compound characters as discussed above incorporate feature extraction using three different features. These features are applied to three different neural networks designed as per the parameters given in Table 8.6. Table 8.6 Neural network parameter settings for hybrid network Parameters Values Number of inputs 35 : Normalized pixel density features 32 : Euclidean distance features 64 : Modified wavelet approximation features Number of hidden layers 1 Number of neurons in hidden layer Equal to the square root of the product of number of inputs and number of outputs Hidden layer activation function hyperbolic tangent sigmoid transfer function Number of neurons in output layer Number of characters in the structural class Output layer activation function Linear Goal Error function mse Maximum number of epoch 300 Training algorithm Levenberg-Marquardt algorithm As seen in Table 8.6, the number of features applied to each neural network in the hybrid network is different. This in turn results in the difference in the number of hidden neurons, which is a function of number of input neurons. Each network generates an 8. Compound Character Recognition using Various Models 152

22 index for the character under recognition. If the indices of any two networks match, the same is used to obtain the recognized character. If the indices of all the networks differ, then the index of the neural network with modified wavelet approximation features is considered as the final output. This is because; the recognition accuracy of this branch is higher as compared to the other two in the hybrid network Results for compound character recognition using hybrid network As discussed and studied earlier, a hybrid network is designed to improve the results over individual feature extraction and classification combination. The performance of the hybrid network for compound character recognition is presented in Table 8.7. It also summarizes the performance of the individual feature extraction and classification combination. The observations quoted in the table indicate that the hybrid network improves the recognition rate over individual network. The time required to recognize a character is approximately 0.15 sec. Table 8.7 Hybrid network performance for compound character recognition Sr. no. Recognition technique Overall recognition rate (%) 1. NN with Euclidean distance features NN with normalized pixel density features NN with Modified wavelet approximation features 4. Hybrid network (majority voting of 1 to 3) The techniques implemented for the recognition of handwritten Marathi compound characters and their performances are summarized in Table Compound Character Recognition using Various Models 153

23 Table 8.8 Summary of results of various systems for compound character recognition Sr. no. Template/Feature generation technique Resize factor Template/Feature size Recognition Technique used Overall Recognition rate (%) 1. Binary templates Convolved binary templates Wavelet approximation templates Modified wavelet approximation templates 16x16 16x x32 32x x16 31x x32 63x x16 8x8 Template matching 32x32 16x x16 31x x32 63x Euclidean distance features 16x16 32 Neural network Normalized pixel density features 70x50 35 Neural network Modified wavelet approximation features 5. Multiple (normalized pixel density, Euclidean distance & Modified wavelet approximation features) 16x16 64 Neural network Differ as Differ as per the Hybrid per the feature extraction network feature technique extraction technique Compound Character Recognition using Various Models 154

24 8.7 Concluding Remarks Handwritten Marathi compound character recognition without separation is studied in this chapter. No such system for handwritten Marathi compound character recognition is found so far to the best of our knowledge. A huge database of compound characters is created. Various recognition techniques were implemented right from template matching to hybrid network in order to improve the performance in a systematic manner. Table 8.9 presents a comparative study of all the three techniques implemented for handwritten Marathi compound character recognition. Table 8.9 Comparative study of various systems for compound character recognition Sr. no. Template/Feature generation technique Resize factor Template/Feature size Recognition Technique used Overall Recognition rate (%) 1. Modified wavelet 32x32 63x63 Template approximation templates matching 2. Modified wavelet 16x16 64 Neural approximation features network 3. Multiple (normalized pixel density, Euclidean distance & Modified wavelet approximation features) Differ as per the feature extraction technique Differ as per the feature extraction technique Hybrid network According to the observations, although the template matching gives acceptable results with modified wavelet features, the time required to recognize a character is large as shown in Table 8.2. Hence neural network is used which improves the accuracy (refer Table 8.9) due to its learning and generalization ability as well as requires quite a less time for recognition (approximately 0.05 sec per character). The hybrid network that combines the outputs from three different neural networks gives the highest recognition accuracy of 97.95%. 8. Compound Character Recognition using Various Models 155

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

Handwritten Hindi Numerals Recognition System

Handwritten Hindi Numerals Recognition System CS365 Project Report Handwritten Hindi Numerals Recognition System Submitted by: Akarshan Sarkar Kritika Singh Project Mentor: Prof. Amitabha Mukerjee 1 Abstract In this project, we consider the problem

More information

Comparative Performance Analysis of Feature(S)- Classifier Combination for Devanagari Optical Character Recognition System

Comparative Performance Analysis of Feature(S)- Classifier Combination for Devanagari Optical Character Recognition System Comparative Performance Analysis of Feature(S)- Classifier Combination for Devanagari Optical Character Recognition System Jasbir Singh Department of Computer Science Punjabi University Patiala, India

More information

Handwritten Script Recognition at Block Level

Handwritten Script Recognition at Block Level Chapter 4 Handwritten Script Recognition at Block Level -------------------------------------------------------------------------------------------------------------------------- Optical character recognition

More information

OCR For Handwritten Marathi Script

OCR For Handwritten Marathi Script International Journal of Scientific & Engineering Research Volume 3, Issue 8, August-2012 1 OCR For Handwritten Marathi Script Mrs.Vinaya. S. Tapkir 1, Mrs.Sushma.D.Shelke 2 1 Maharashtra Academy Of Engineering,

More information

CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS

CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS CHAPTER 4 CLASSIFICATION WITH RADIAL BASIS AND PROBABILISTIC NEURAL NETWORKS 4.1 Introduction Optical character recognition is one of

More information

HANDWRITTEN GURMUKHI CHARACTER RECOGNITION USING WAVELET TRANSFORMS

HANDWRITTEN GURMUKHI CHARACTER RECOGNITION USING WAVELET TRANSFORMS International Journal of Electronics, Communication & Instrumentation Engineering Research and Development (IJECIERD) ISSN 2249-684X Vol.2, Issue 3 Sep 2012 27-37 TJPRC Pvt. Ltd., HANDWRITTEN GURMUKHI

More information

Fine Classification of Unconstrained Handwritten Persian/Arabic Numerals by Removing Confusion amongst Similar Classes

Fine Classification of Unconstrained Handwritten Persian/Arabic Numerals by Removing Confusion amongst Similar Classes 2009 10th International Conference on Document Analysis and Recognition Fine Classification of Unconstrained Handwritten Persian/Arabic Numerals by Removing Confusion amongst Similar Classes Alireza Alaei

More information

SEVERAL METHODS OF FEATURE EXTRACTION TO HELP IN OPTICAL CHARACTER RECOGNITION

SEVERAL METHODS OF FEATURE EXTRACTION TO HELP IN OPTICAL CHARACTER RECOGNITION SEVERAL METHODS OF FEATURE EXTRACTION TO HELP IN OPTICAL CHARACTER RECOGNITION Binod Kumar Prasad * * Bengal College of Engineering and Technology, Durgapur, W.B., India. Rajdeep Kundu 2 2 Bengal College

More information

Structural Feature Extraction to recognize some of the Offline Isolated Handwritten Gujarati Characters using Decision Tree Classifier

Structural Feature Extraction to recognize some of the Offline Isolated Handwritten Gujarati Characters using Decision Tree Classifier Structural Feature Extraction to recognize some of the Offline Isolated Handwritten Gujarati Characters using Decision Tree Classifier Hetal R. Thaker Atmiya Institute of Technology & science, Kalawad

More information

LITERATURE REVIEW. For Indian languages most of research work is performed firstly on Devnagari script and secondly on Bangla script.

LITERATURE REVIEW. For Indian languages most of research work is performed firstly on Devnagari script and secondly on Bangla script. LITERATURE REVIEW For Indian languages most of research work is performed firstly on Devnagari script and secondly on Bangla script. The study of recognition for handwritten Devanagari compound character

More information

Isolated Curved Gurmukhi Character Recognition Using Projection of Gradient

Isolated Curved Gurmukhi Character Recognition Using Projection of Gradient International Journal of Computational Intelligence Research ISSN 0973-1873 Volume 13, Number 6 (2017), pp. 1387-1396 Research India Publications http://www.ripublication.com Isolated Curved Gurmukhi Character

More information

A System for Joining and Recognition of Broken Bangla Numerals for Indian Postal Automation

A System for Joining and Recognition of Broken Bangla Numerals for Indian Postal Automation A System for Joining and Recognition of Broken Bangla Numerals for Indian Postal Automation K. Roy, U. Pal and B. B. Chaudhuri CVPR Unit; Indian Statistical Institute, Kolkata-108; India umapada@isical.ac.in

More information

COMBINING NEURAL NETWORKS FOR SKIN DETECTION

COMBINING NEURAL NETWORKS FOR SKIN DETECTION COMBINING NEURAL NETWORKS FOR SKIN DETECTION Chelsia Amy Doukim 1, Jamal Ahmad Dargham 1, Ali Chekima 1 and Sigeru Omatu 2 1 School of Engineering and Information Technology, Universiti Malaysia Sabah,

More information

Spotting Words in Latin, Devanagari and Arabic Scripts

Spotting Words in Latin, Devanagari and Arabic Scripts Spotting Words in Latin, Devanagari and Arabic Scripts Sargur N. Srihari, Harish Srinivasan, Chen Huang and Shravya Shetty {srihari,hs32,chuang5,sshetty}@cedar.buffalo.edu Center of Excellence for Document

More information

Texture Analysis of Painted Strokes 1) Martin Lettner, Paul Kammerer, Robert Sablatnig

Texture Analysis of Painted Strokes 1) Martin Lettner, Paul Kammerer, Robert Sablatnig Texture Analysis of Painted Strokes 1) Martin Lettner, Paul Kammerer, Robert Sablatnig Vienna University of Technology, Institute of Computer Aided Automation, Pattern Recognition and Image Processing

More information

Neural Network Neurons

Neural Network Neurons Neural Networks Neural Network Neurons 1 Receives n inputs (plus a bias term) Multiplies each input by its weight Applies activation function to the sum of results Outputs result Activation Functions Given

More information

CHAPTER VI BACK PROPAGATION ALGORITHM

CHAPTER VI BACK PROPAGATION ALGORITHM 6.1 Introduction CHAPTER VI BACK PROPAGATION ALGORITHM In the previous chapter, we analysed that multiple layer perceptrons are effectively applied to handle tricky problems if trained with a vastly accepted

More information

Radial Basis Function Neural Network Classifier

Radial Basis Function Neural Network Classifier Recognition of Unconstrained Handwritten Numerals by a Radial Basis Function Neural Network Classifier Hwang, Young-Sup and Bang, Sung-Yang Department of Computer Science & Engineering Pohang University

More information

Visual object classification by sparse convolutional neural networks

Visual object classification by sparse convolutional neural networks Visual object classification by sparse convolutional neural networks Alexander Gepperth 1 1- Ruhr-Universität Bochum - Institute for Neural Dynamics Universitätsstraße 150, 44801 Bochum - Germany Abstract.

More information

Handwritten Gurumukhi Character Recognition by using Recurrent Neural Network

Handwritten Gurumukhi Character Recognition by using Recurrent Neural Network 139 Handwritten Gurumukhi Character Recognition by using Recurrent Neural Network Harmit Kaur 1, Simpel Rani 2 1 M. Tech. Research Scholar (Department of Computer Science & Engineering), Yadavindra College

More information

Optical Character Recognition (OCR) for Printed Devnagari Script Using Artificial Neural Network

Optical Character Recognition (OCR) for Printed Devnagari Script Using Artificial Neural Network International Journal of Computer Science & Communication Vol. 1, No. 1, January-June 2010, pp. 91-95 Optical Character Recognition (OCR) for Printed Devnagari Script Using Artificial Neural Network Raghuraj

More information

MOMENT AND DENSITY BASED HADWRITTEN MARATHI NUMERAL RECOGNITION

MOMENT AND DENSITY BASED HADWRITTEN MARATHI NUMERAL RECOGNITION MOMENT AND DENSITY BASED HADWRITTEN MARATHI NUMERAL RECOGNITION S. M. Mali Department of Computer Science, MAEER S Arts, Commerce and Science College, Pune Shankarmali007@gmail.com Abstract In this paper,

More information

Equation to LaTeX. Abhinav Rastogi, Sevy Harris. I. Introduction. Segmentation.

Equation to LaTeX. Abhinav Rastogi, Sevy Harris. I. Introduction. Segmentation. Equation to LaTeX Abhinav Rastogi, Sevy Harris {arastogi,sharris5}@stanford.edu I. Introduction Copying equations from a pdf file to a LaTeX document can be time consuming because there is no easy way

More information

TERM PAPER ON The Compressive Sensing Based on Biorthogonal Wavelet Basis

TERM PAPER ON The Compressive Sensing Based on Biorthogonal Wavelet Basis TERM PAPER ON The Compressive Sensing Based on Biorthogonal Wavelet Basis Submitted By: Amrita Mishra 11104163 Manoj C 11104059 Under the Guidance of Dr. Sumana Gupta Professor Department of Electrical

More information

CHAPTER 6 DETECTION OF MASS USING NOVEL SEGMENTATION, GLCM AND NEURAL NETWORKS

CHAPTER 6 DETECTION OF MASS USING NOVEL SEGMENTATION, GLCM AND NEURAL NETWORKS 130 CHAPTER 6 DETECTION OF MASS USING NOVEL SEGMENTATION, GLCM AND NEURAL NETWORKS A mass is defined as a space-occupying lesion seen in more than one projection and it is described by its shapes and margin

More information

Handwritten Devanagari Character Recognition Model Using Neural Network

Handwritten Devanagari Character Recognition Model Using Neural Network Handwritten Devanagari Character Recognition Model Using Neural Network Gaurav Jaiswal M.Sc. (Computer Science) Department of Computer Science Banaras Hindu University, Varanasi. India gauravjais88@gmail.com

More information

CHAPTER 3 WAVELET DECOMPOSITION USING HAAR WAVELET

CHAPTER 3 WAVELET DECOMPOSITION USING HAAR WAVELET 69 CHAPTER 3 WAVELET DECOMPOSITION USING HAAR WAVELET 3.1 WAVELET Wavelet as a subject is highly interdisciplinary and it draws in crucial ways on ideas from the outside world. The working of wavelet in

More information

FRAGMENTATION OF HANDWRITTEN TOUCHING CHARACTERS IN DEVANAGARI SCRIPT

FRAGMENTATION OF HANDWRITTEN TOUCHING CHARACTERS IN DEVANAGARI SCRIPT International Journal of Information Technology, Modeling and Computing (IJITMC) Vol. 2, No. 1, February 2014 FRAGMENTATION OF HANDWRITTEN TOUCHING CHARACTERS IN DEVANAGARI SCRIPT Shuchi Kapoor 1 and Vivek

More information

Machine Learning 13. week

Machine Learning 13. week Machine Learning 13. week Deep Learning Convolutional Neural Network Recurrent Neural Network 1 Why Deep Learning is so Popular? 1. Increase in the amount of data Thanks to the Internet, huge amount of

More information

Convolution Neural Networks for Chinese Handwriting Recognition

Convolution Neural Networks for Chinese Handwriting Recognition Convolution Neural Networks for Chinese Handwriting Recognition Xu Chen Stanford University 450 Serra Mall, Stanford, CA 94305 xchen91@stanford.edu Abstract Convolutional neural networks have been proven

More information

DETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT

DETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT DETECTION OF SMOOTH TEXTURE IN FACIAL IMAGES FOR THE EVALUATION OF UNNATURAL CONTRAST ENHANCEMENT 1 NUR HALILAH BINTI ISMAIL, 2 SOONG-DER CHEN 1, 2 Department of Graphics and Multimedia, College of Information

More information

Handwritten character and word recognition using their geometrical features through neural networks

Handwritten character and word recognition using their geometrical features through neural networks Handwritten character and word recognition using their geometrical features through neural networks Sudarshan Sawant 1, Prof. Seema Baji 2 1 Student, Department of electronics and Tele-communications,

More information

Neural Network based textural labeling of images in multimedia applications

Neural Network based textural labeling of images in multimedia applications Neural Network based textural labeling of images in multimedia applications S.A. Karkanis +, G.D. Magoulas +, and D.A. Karras ++ + University of Athens, Dept. of Informatics, Typa Build., Panepistimiopolis,

More information

A Fast Personal Palm print Authentication based on 3D-Multi Wavelet Transformation

A Fast Personal Palm print Authentication based on 3D-Multi Wavelet Transformation A Fast Personal Palm print Authentication based on 3D-Multi Wavelet Transformation * A. H. M. Al-Helali, * W. A. Mahmmoud, and * H. A. Ali * Al- Isra Private University Email: adnan_hadi@yahoo.com Abstract:

More information

Chapter Review of HCR

Chapter Review of HCR Chapter 3 [3]Literature Review The survey of literature on character recognition showed that some of the researchers have worked based on application requirements like postal code identification [118],

More information

ISSN (ONLINE): , VOLUME-3, ISSUE-1,

ISSN (ONLINE): , VOLUME-3, ISSUE-1, PERFORMANCE ANALYSIS OF LOSSLESS COMPRESSION TECHNIQUES TO INVESTIGATE THE OPTIMUM IMAGE COMPRESSION TECHNIQUE Dr. S. Swapna Rani Associate Professor, ECE Department M.V.S.R Engineering College, Nadergul,

More information

NOVATEUR PUBLICATIONS INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY [IJIERT] ISSN: VOLUME 5, ISSUE

NOVATEUR PUBLICATIONS INTERNATIONAL JOURNAL OF INNOVATIONS IN ENGINEERING RESEARCH AND TECHNOLOGY [IJIERT] ISSN: VOLUME 5, ISSUE OPTICAL HANDWRITTEN DEVNAGARI CHARACTER RECOGNITION USING ARTIFICIAL NEURAL NETWORK APPROACH JYOTI A.PATIL Ashokrao Mane Group of Institution, Vathar Tarf Vadgaon, India. DR. SANJAY R. PATIL Ashokrao Mane

More information

Segmentation and Grouping

Segmentation and Grouping Segmentation and Grouping How and what do we see? Fundamental Problems ' Focus of attention, or grouping ' What subsets of pixels do we consider as possible objects? ' All connected subsets? ' Representation

More information

Image Compression: An Artificial Neural Network Approach

Image Compression: An Artificial Neural Network Approach Image Compression: An Artificial Neural Network Approach Anjana B 1, Mrs Shreeja R 2 1 Department of Computer Science and Engineering, Calicut University, Kuttippuram 2 Department of Computer Science and

More information

Image Classification using Fast Learning Convolutional Neural Networks

Image Classification using Fast Learning Convolutional Neural Networks , pp.50-55 http://dx.doi.org/10.14257/astl.2015.113.11 Image Classification using Fast Learning Convolutional Neural Networks Keonhee Lee 1 and Dong-Chul Park 2 1 Software Device Research Center Korea

More information

4. RESULTS AND COMPARASIONS

4. RESULTS AND COMPARASIONS 80 4. RESULTS AND COMPARASIONS 4.1 Introduction The previous chapter describes the methodology of the proposed algorithm. In this chapter we discuss the results obtained by the proposed algorithms. We

More information

Character Recognition

Character Recognition Character Recognition 5.1 INTRODUCTION Recognition is one of the important steps in image processing. There are different methods such as Histogram method, Hough transformation, Neural computing approaches

More information

International Journal of Scientific Research & Engineering Trends Volume 4, Issue 6, Nov-Dec-2018, ISSN (Online): X

International Journal of Scientific Research & Engineering Trends Volume 4, Issue 6, Nov-Dec-2018, ISSN (Online): X Analysis about Classification Techniques on Categorical Data in Data Mining Assistant Professor P. Meena Department of Computer Science Adhiyaman Arts and Science College for Women Uthangarai, Krishnagiri,

More information

Image Processing. Image Features

Image Processing. Image Features Image Processing Image Features Preliminaries 2 What are Image Features? Anything. What they are used for? Some statements about image fragments (patches) recognition Search for similar patches matching

More information

Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers

Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers Traffic Signs Recognition using HP and HOG Descriptors Combined to MLP and SVM Classifiers A. Salhi, B. Minaoui, M. Fakir, H. Chakib, H. Grimech Faculty of science and Technology Sultan Moulay Slimane

More information

Edge and local feature detection - 2. Importance of edge detection in computer vision

Edge and local feature detection - 2. Importance of edge detection in computer vision Edge and local feature detection Gradient based edge detection Edge detection by function fitting Second derivative edge detectors Edge linking and the construction of the chain graph Edge and local feature

More information

2. LITERATURE REVIEW

2. LITERATURE REVIEW 2. LITERATURE REVIEW CBIR has come long way before 1990 and very little papers have been published at that time, however the number of papers published since 1997 is increasing. There are many CBIR algorithms

More information

Opportunities and Challenges of Handwritten Sanskrit Character Recognition System

Opportunities and Challenges of Handwritten Sanskrit Character Recognition System Opportunities and Challenges of Handwritten System Shailendra Kumar Singh Research Scholar, CSE Department SLIET Longowal, Sangrur, Punjab, India Sks.it2012@gmail.com Manoj Kumar Sachan Assosiate Professor,

More information

In this assignment, we investigated the use of neural networks for supervised classification

In this assignment, we investigated the use of neural networks for supervised classification Paul Couchman Fabien Imbault Ronan Tigreat Gorka Urchegui Tellechea Classification assignment (group 6) Image processing MSc Embedded Systems March 2003 Classification includes a broad range of decision-theoric

More information

Implementation Of Neural Network Based Script Recognition

Implementation Of Neural Network Based Script Recognition Implementation Of Neural Network Based Script Recognition Manisha Kumavat, Alwin Anuse Department of Electronics and Telecommunication Maharashtra Institute of Technology, Pune Abstract A variety of different

More information

Online Bangla Handwriting Recognition System

Online Bangla Handwriting Recognition System 1 Online Bangla Handwriting Recognition System K. Roy Dept. of Comp. Sc. West Bengal University of Technology, BF 142, Saltlake, Kolkata-64, India N. Sharma, T. Pal and U. Pal Computer Vision and Pattern

More information

Word-wise Hand-written Script Separation for Indian Postal automation

Word-wise Hand-written Script Separation for Indian Postal automation Word-wise Hand-written Script Separation for Indian Postal automation K. Roy U. Pal Dept. of Comp. Sc. & Engg. West Bengal University of Technology, Sector 1, Saltlake City, Kolkata-64, India Abstract

More information

Supervised Learning in Neural Networks (Part 2)

Supervised Learning in Neural Networks (Part 2) Supervised Learning in Neural Networks (Part 2) Multilayer neural networks (back-propagation training algorithm) The input signals are propagated in a forward direction on a layer-bylayer basis. Learning

More information

Why MultiLayer Perceptron/Neural Network? Objective: Attributes:

Why MultiLayer Perceptron/Neural Network? Objective: Attributes: Why MultiLayer Perceptron/Neural Network? Neural networks, with their remarkable ability to derive meaning from complicated or imprecise data, can be used to extract patterns and detect trends that are

More information

CMU Lecture 18: Deep learning and Vision: Convolutional neural networks. Teacher: Gianni A. Di Caro

CMU Lecture 18: Deep learning and Vision: Convolutional neural networks. Teacher: Gianni A. Di Caro CMU 15-781 Lecture 18: Deep learning and Vision: Convolutional neural networks Teacher: Gianni A. Di Caro DEEP, SHALLOW, CONNECTED, SPARSE? Fully connected multi-layer feed-forward perceptrons: More powerful

More information

THE MNIST DATABASE of handwritten digits Yann LeCun, Courant Institute, NYU Corinna Cortes, Google Labs, New York

THE MNIST DATABASE of handwritten digits Yann LeCun, Courant Institute, NYU Corinna Cortes, Google Labs, New York THE MNIST DATABASE of handwritten digits Yann LeCun, Courant Institute, NYU Corinna Cortes, Google Labs, New York The MNIST database of handwritten digits, available from this page, has a training set

More information

A Generalized Method to Solve Text-Based CAPTCHAs

A Generalized Method to Solve Text-Based CAPTCHAs A Generalized Method to Solve Text-Based CAPTCHAs Jason Ma, Bilal Badaoui, Emile Chamoun December 11, 2009 1 Abstract We present work in progress on the automated solving of text-based CAPTCHAs. Our method

More information

DEVANAGARI SCRIPT SEPARATION AND RECOGNITION USING MORPHOLOGICAL OPERATIONS AND OPTIMIZED FEATURE EXTRACTION METHODS

DEVANAGARI SCRIPT SEPARATION AND RECOGNITION USING MORPHOLOGICAL OPERATIONS AND OPTIMIZED FEATURE EXTRACTION METHODS DEVANAGARI SCRIPT SEPARATION AND RECOGNITION USING MORPHOLOGICAL OPERATIONS AND OPTIMIZED FEATURE EXTRACTION METHODS Sushilkumar N. Holambe Dr. Ulhas B. Shinde Shrikant D. Mali Persuing PhD at Principal

More information

The Detection of Faces in Color Images: EE368 Project Report

The Detection of Faces in Color Images: EE368 Project Report The Detection of Faces in Color Images: EE368 Project Report Angela Chau, Ezinne Oji, Jeff Walters Dept. of Electrical Engineering Stanford University Stanford, CA 9435 angichau,ezinne,jwalt@stanford.edu

More information

Image Normalization and Preprocessing for Gujarati Character Recognition

Image Normalization and Preprocessing for Gujarati Character Recognition 334 Image Normalization and Preprocessing for Gujarati Character Recognition Jayashree Rajesh Prasad Department of Computer Engineering, Sinhgad College of Engineering, University of Pune, Pune, Mahaashtra

More information

Recognition of Handwritten Digits using Machine Learning Techniques

Recognition of Handwritten Digits using Machine Learning Techniques Recognition of Handwritten Digits using Machine Learning Techniques Shobhit Srivastava #1, Sanjana Kalani #2,Umme Hani #3, Sayak Chakraborty #4 Department of Computer Science and Engineering Dayananda

More information

Recognizing Handwritten Digits Using the LLE Algorithm with Back Propagation

Recognizing Handwritten Digits Using the LLE Algorithm with Back Propagation Recognizing Handwritten Digits Using the LLE Algorithm with Back Propagation Lori Cillo, Attebury Honors Program Dr. Rajan Alex, Mentor West Texas A&M University Canyon, Texas 1 ABSTRACT. This work is

More information

Journal of Asian Scientific Research FEATURES COMPOSITION FOR PROFICIENT AND REAL TIME RETRIEVAL IN CBIR SYSTEM. Tohid Sedghi

Journal of Asian Scientific Research FEATURES COMPOSITION FOR PROFICIENT AND REAL TIME RETRIEVAL IN CBIR SYSTEM. Tohid Sedghi Journal of Asian Scientific Research, 013, 3(1):68-74 Journal of Asian Scientific Research journal homepage: http://aessweb.com/journal-detail.php?id=5003 FEATURES COMPOSTON FOR PROFCENT AND REAL TME RETREVAL

More information

Recognition of Off-Line Handwritten Devnagari Characters Using Quadratic Classifier

Recognition of Off-Line Handwritten Devnagari Characters Using Quadratic Classifier Recognition of Off-Line Handwritten Devnagari Characters Using Quadratic Classifier N. Sharma, U. Pal*, F. Kimura**, and S. Pal Computer Vision and Pattern Recognition Unit, Indian Statistical Institute

More information

Review of Automatic Handwritten Kannada Character Recognition Technique Using Neural Network

Review of Automatic Handwritten Kannada Character Recognition Technique Using Neural Network Review of Automatic Handwritten Kannada Character Recognition Technique Using Neural Network 1 Mukesh Kumar, 2 Dr.Jeeetendra Sheethlani 1 Department of Computer Science SSSUTMS, Sehore Abstract Data processing

More information

A COMPARISON OF WAVELET-BASED AND RIDGELET- BASED TEXTURE CLASSIFICATION OF TISSUES IN COMPUTED TOMOGRAPHY

A COMPARISON OF WAVELET-BASED AND RIDGELET- BASED TEXTURE CLASSIFICATION OF TISSUES IN COMPUTED TOMOGRAPHY A COMPARISON OF WAVELET-BASED AND RIDGELET- BASED TEXTURE CLASSIFICATION OF TISSUES IN COMPUTED TOMOGRAPHY Lindsay Semler Lucia Dettori Intelligent Multimedia Processing Laboratory School of Computer Scienve,

More information

Assignment 2. Classification and Regression using Linear Networks, Multilayer Perceptron Networks, and Radial Basis Functions

Assignment 2. Classification and Regression using Linear Networks, Multilayer Perceptron Networks, and Radial Basis Functions ENEE 739Q: STATISTICAL AND NEURAL PATTERN RECOGNITION Spring 2002 Assignment 2 Classification and Regression using Linear Networks, Multilayer Perceptron Networks, and Radial Basis Functions Aravind Sundaresan

More information

A Brief Study of Feature Extraction and Classification Methods Used for Character Recognition of Brahmi Northern Indian Scripts

A Brief Study of Feature Extraction and Classification Methods Used for Character Recognition of Brahmi Northern Indian Scripts 25 A Brief Study of Feature Extraction and Classification Methods Used for Character Recognition of Brahmi Northern Indian Scripts Rohit Sachdeva, Asstt. Prof., Computer Science Department, Multani Mal

More information

Complementary Features Combined in a MLP-based System to Recognize Handwritten Devnagari Character

Complementary Features Combined in a MLP-based System to Recognize Handwritten Devnagari Character Journal of Information Hiding and Multimedia Signal Processing 2011 ISSN 2073-4212 Ubiquitous International Volume 2, Number 1, January 2011 Complementary Features Combined in a MLP-based System to Recognize

More information

CHAPTER 4 STOCK PRICE PREDICTION USING MODIFIED K-NEAREST NEIGHBOR (MKNN) ALGORITHM

CHAPTER 4 STOCK PRICE PREDICTION USING MODIFIED K-NEAREST NEIGHBOR (MKNN) ALGORITHM CHAPTER 4 STOCK PRICE PREDICTION USING MODIFIED K-NEAREST NEIGHBOR (MKNN) ALGORITHM 4.1 Introduction Nowadays money investment in stock market gains major attention because of its dynamic nature. So the

More information

Chapter 6 CLASSIFICATION ALGORITHMS FOR DETECTION OF ABNORMALITIES IN MAMMOGRAM IMAGES

Chapter 6 CLASSIFICATION ALGORITHMS FOR DETECTION OF ABNORMALITIES IN MAMMOGRAM IMAGES CLASSIFICATION ALGORITHMS FOR DETECTION OF ABNORMALITIES IN MAMMOGRAM IMAGES The two deciding factors of an efficient system for the detection of abnormalities are the nature and type of features extracted

More information

Convolutional Neural Networks

Convolutional Neural Networks Lecturer: Barnabas Poczos Introduction to Machine Learning (Lecture Notes) Convolutional Neural Networks Disclaimer: These notes have not been subjected to the usual scrutiny reserved for formal publications.

More information

One Dim~nsional Representation Of Two Dimensional Information For HMM Based Handwritten Recognition

One Dim~nsional Representation Of Two Dimensional Information For HMM Based Handwritten Recognition One Dim~nsional Representation Of Two Dimensional Information For HMM Based Handwritten Recognition Nafiz Arica Dept. of Computer Engineering, Middle East Technical University, Ankara,Turkey nafiz@ceng.metu.edu.

More information

Tutorial 5. Jun Xu, Teaching Asistant March 2, COMP4134 Biometrics Authentication

Tutorial 5. Jun Xu, Teaching Asistant March 2, COMP4134 Biometrics Authentication Tutorial 5 Jun Xu, Teaching Asistant nankaimathxujun@gmail.com COMP4134 Biometrics Authentication March 2, 2017 Table of Contents Problems Problem 1: Answer The Questions Problem 2: Indeterminate Region

More information

ADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N.

ADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N. ADVANCED IMAGE PROCESSING METHODS FOR ULTRASONIC NDE RESEARCH C. H. Chen, University of Massachusetts Dartmouth, N. Dartmouth, MA USA Abstract: The significant progress in ultrasonic NDE systems has now

More information

Hand Written Character Recognition using VNP based Segmentation and Artificial Neural Network

Hand Written Character Recognition using VNP based Segmentation and Artificial Neural Network International Journal of Emerging Engineering Research and Technology Volume 4, Issue 6, June 2016, PP 38-46 ISSN 2349-4395 (Print) & ISSN 2349-4409 (Online) Hand Written Character Recognition using VNP

More information

APPLICATION OF RECIRCULATION NEURAL NETWORK AND PRINCIPAL COMPONENT ANALYSIS FOR FACE RECOGNITION

APPLICATION OF RECIRCULATION NEURAL NETWORK AND PRINCIPAL COMPONENT ANALYSIS FOR FACE RECOGNITION APPLICATION OF RECIRCULATION NEURAL NETWORK AND PRINCIPAL COMPONENT ANALYSIS FOR FACE RECOGNITION Dmitr Brliuk and Valer Starovoitov Institute of Engineering Cbernetics, Laborator of Image Processing and

More information

An Algorithm For Training Multilayer Perceptron (MLP) For Image Reconstruction Using Neural Network Without Overfitting.

An Algorithm For Training Multilayer Perceptron (MLP) For Image Reconstruction Using Neural Network Without Overfitting. An Algorithm For Training Multilayer Perceptron (MLP) For Image Reconstruction Using Neural Network Without Overfitting. Mohammad Mahmudul Alam Mia, Shovasis Kumar Biswas, Monalisa Chowdhury Urmi, Abubakar

More information

INTERNATIONAL JOURNAL OF PURE AND APPLIED RESEARCH IN ENGINEERING AND TECHNOLOGY

INTERNATIONAL JOURNAL OF PURE AND APPLIED RESEARCH IN ENGINEERING AND TECHNOLOGY INTERNATIONAL JOURNAL OF PURE AND APPLIED RESEARCH IN ENGINEERING AND TECHNOLOGY A PATH FOR HORIZING YOUR INNOVATIVE WORK HANDWRITTEN DEVANAGARI CHARACTERS RECOGNITION THROUGH SEGMENTATION AND ARTIFICIAL

More information

Solar Radiation Data Modeling with a Novel Surface Fitting Approach

Solar Radiation Data Modeling with a Novel Surface Fitting Approach Solar Radiation Data Modeling with a Novel Surface Fitting Approach F. Onur Hocao glu, Ömer Nezih Gerek, Mehmet Kurban Anadolu University, Dept. of Electrical and Electronics Eng., Eskisehir, Turkey {fohocaoglu,ongerek,mkurban}

More information

MULTIVARIATE TEXTURE DISCRIMINATION USING A PRINCIPAL GEODESIC CLASSIFIER

MULTIVARIATE TEXTURE DISCRIMINATION USING A PRINCIPAL GEODESIC CLASSIFIER MULTIVARIATE TEXTURE DISCRIMINATION USING A PRINCIPAL GEODESIC CLASSIFIER A.Shabbir 1, 2 and G.Verdoolaege 1, 3 1 Department of Applied Physics, Ghent University, B-9000 Ghent, Belgium 2 Max Planck Institute

More information

A Technique for Offline Handwritten Character Recognition

A Technique for Offline Handwritten Character Recognition A Technique for Offline Handwritten Character Recognition 1 Shilpy Bansal, 2 Mamta Garg, 3 Munish Kumar 1 Lecturer, Department of Computer Science Engineering, BMSCET, Muktsar, Punjab 2 Assistant Professor,

More information

A Survey of Problems of Overlapped Handwritten Characters in Recognition process for Gurmukhi Script

A Survey of Problems of Overlapped Handwritten Characters in Recognition process for Gurmukhi Script A Survey of Problems of Overlapped Handwritten Characters in Recognition process for Gurmukhi Script Arwinder Kaur 1, Ashok Kumar Bathla 2 1 M. Tech. Student, CE Dept., 2 Assistant Professor, CE Dept.,

More information

Linear Discriminant Analysis for 3D Face Recognition System

Linear Discriminant Analysis for 3D Face Recognition System Linear Discriminant Analysis for 3D Face Recognition System 3.1 Introduction Face recognition and verification have been at the top of the research agenda of the computer vision community in recent times.

More information

1. Introduction. 2. Motivation and Problem Definition. Volume 8 Issue 2, February Susmita Mohapatra

1. Introduction. 2. Motivation and Problem Definition. Volume 8 Issue 2, February Susmita Mohapatra Pattern Recall Analysis of the Hopfield Neural Network with a Genetic Algorithm Susmita Mohapatra Department of Computer Science, Utkal University, India Abstract: This paper is focused on the implementation

More information

CS229 Final Project: Predicting Expected Response Times

CS229 Final Project: Predicting Expected  Response Times CS229 Final Project: Predicting Expected Email Response Times Laura Cruz-Albrecht (lcruzalb), Kevin Khieu (kkhieu) December 15, 2017 1 Introduction Each day, countless emails are sent out, yet the time

More information

Neural Networks. CE-725: Statistical Pattern Recognition Sharif University of Technology Spring Soleymani

Neural Networks. CE-725: Statistical Pattern Recognition Sharif University of Technology Spring Soleymani Neural Networks CE-725: Statistical Pattern Recognition Sharif University of Technology Spring 2013 Soleymani Outline Biological and artificial neural networks Feed-forward neural networks Single layer

More information

Indian Multi-Script Full Pin-code String Recognition for Postal Automation

Indian Multi-Script Full Pin-code String Recognition for Postal Automation 2009 10th International Conference on Document Analysis and Recognition Indian Multi-Script Full Pin-code String Recognition for Postal Automation U. Pal 1, R. K. Roy 1, K. Roy 2 and F. Kimura 3 1 Computer

More information

Off-line Recognition of Hand-written Bengali Numerals using Morphological Features

Off-line Recognition of Hand-written Bengali Numerals using Morphological Features Off-line Recognition of Hand-written Bengali Numerals using Morphological Features Pulak Purkait and Bhabatosh Chanda ECSU, Indian Statistical Institute, Kolkata, India {pulak r, chanda}@isical.ac.in Abstract

More information

Word-wise Script Identification from Video Frames

Word-wise Script Identification from Video Frames Word-wise Script Identification from Video Frames Author Sharma, Nabin, Chanda, Sukalpa, Pal, Umapada, Blumenstein, Michael Published 2013 Conference Title Proceedings 12th International Conference on

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

Edge Histogram Descriptor, Geometric Moment and Sobel Edge Detector Combined Features Based Object Recognition and Retrieval System

Edge Histogram Descriptor, Geometric Moment and Sobel Edge Detector Combined Features Based Object Recognition and Retrieval System Edge Histogram Descriptor, Geometric Moment and Sobel Edge Detector Combined Features Based Object Recognition and Retrieval System Neetesh Prajapati M. Tech Scholar VNS college,bhopal Amit Kumar Nandanwar

More information

HCR Using K-Means Clustering Algorithm

HCR Using K-Means Clustering Algorithm HCR Using K-Means Clustering Algorithm Meha Mathur 1, Anil Saroliya 2 Amity School of Engineering & Technology Amity University Rajasthan, India Abstract: Hindi is a national language of India, there are

More information

Multi-Layer Perceptron Network For Handwritting English Character Recoginition

Multi-Layer Perceptron Network For Handwritting English Character Recoginition Multi-Layer Perceptron Network For Handwritting English Character Recoginition 1 Mohit Mittal, 2 Tarun Bhalla 1,2 Anand College of Engg & Mgmt., Kapurthala, Punjab, India Abstract Handwriting recognition

More information

Rapid Natural Scene Text Segmentation

Rapid Natural Scene Text Segmentation Rapid Natural Scene Text Segmentation Ben Newhouse, Stanford University December 10, 2009 1 Abstract A new algorithm was developed to segment text from an image by classifying images according to the gradient

More information

PCA-based Offline Handwritten Character Recognition System

PCA-based Offline Handwritten Character Recognition System Smart Computing Review, vol. 3, no. 5, October 2013 346 Smart Computing Review PCA-based Offline Handwritten Character Recognition System Munish Kumar 1, M. K. Jindal 2, and R. K. Sharma 3 1 Computer Science

More information

Handwritten English Alphabet Recognition Using Bigram Cost Chengshu (Eric) Li Fall 2015, CS229, Stanford University

Handwritten English Alphabet Recognition Using Bigram Cost Chengshu (Eric) Li Fall 2015, CS229, Stanford University Handwritten English Alphabet Recognition Using Bigram Cost Chengshu (Eric) Li chengshu@stanford.edu Fall 2015, CS229, Stanford University Abstract: This paper describes a new approach to handwritten English

More information

Seminar. Topic: Object and character Recognition

Seminar. Topic: Object and character Recognition Seminar Topic: Object and character Recognition Tse Ngang Akumawah Lehrstuhl für Praktische Informatik 3 Table of content What's OCR? Areas covered in OCR Procedure Where does clustering come in Neural

More information

Reversible Wavelets for Embedded Image Compression. Sri Rama Prasanna Pavani Electrical and Computer Engineering, CU Boulder

Reversible Wavelets for Embedded Image Compression. Sri Rama Prasanna Pavani Electrical and Computer Engineering, CU Boulder Reversible Wavelets for Embedded Image Compression Sri Rama Prasanna Pavani Electrical and Computer Engineering, CU Boulder pavani@colorado.edu APPM 7400 - Wavelets and Imaging Prof. Gregory Beylkin -

More information