Karamjit Bhatia - Academia.edu (original) (raw)
Papers by Karamjit Bhatia
Int. J. Comput. Vis. Image Process., 2021
A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike c... more A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike complex moments like a tool to describe the characteristics of a character. Further, an algorithm for selecting the features is employed to identify the substantial image moments from the extracted moments, as the extracted moments may have some insignificant ones. Insignificant moments can increase the computational time and can also degrade the classification accuracy. Thus, the objectives of the study are twofold: (1) to find the important Zernike moments by employing the Genetic algorithm (GA) and (2) the classification of each character is performed using neural network. This way, the performance of the proposed technique is evaluated on two parameters (i.e., speed and recognition accuracy). Further, the efficacy of GA for selecting the moment features is assessed, and the efficacy of selected Zernike complex moments using GA is analyzed for handwritten Hindi characters. Here, the au...
Offline handwritten signature verification is a very challenging area of research as the handwrit... more Offline handwritten signature verification is a very challenging area of research as the handwriting of two people may bear similarity whereas handwriting of a person may vary at different times. The accuracy of handwritten signature verification system depends on the classifier system and the way of feature extraction. Keeping this point of view, four types of hybrid feature sets and three types of classifiers specifically support vector machine with polynomial kernel, support vector machine with quadratic kernel and decision tree are investigated for writer-independent offline handwritten signature verification in the present work. To obtain hybrid feature sets, local oriented statistical information booster, discrete wavelet transform, and histogram of oriented gradient feature descriptors are extracted and are coupled with each other. To create multiple classifier system, the training set is partitioned into subsets and these training subsets are used to train the classifiers of...
International Journal of Tomography and Simulation, 2019
Feature selection mechanism plays a vital role in attaining higher recognition rate in machine ba... more Feature selection mechanism plays a vital role in attaining higher recognition rate in machine based character recognition systems. Zernike moments and Zernike complex moments are found to be very beneficial as a basic tool to generate shape descriptors of a digital image, as they are capable of representing image features efficiently. These moments have been successfully utilized to solve some of the image processing problems and possess the property of rotation invariance which is a desirable aspect in many applications requiring a proficient offline character recognition system. This paper is devoted for inspecting the efficacy of Zernike moments and Zernike complex moments for offline Hindi numerals recognition. Feature vectors consisting of Zernike moments and Zernike complex moments are supplied to the resilient backpropagation neural network for training and different strategies are considered for the performance evaluation of these moments in the study. Character recognition...
Int. J. Comput. Vis. Image Process., 2021
A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike c... more A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike complex moments like a tool to describe the characteristics of a character. Further, an algorithm for selecting the features is employed to identify the substantial image moments from the extracted moments, as the extracted moments may have some insignificant ones. Insignificant moments can increase the computational time and can also degrade the classification accuracy. Thus, the objectives of the study are twofold: (1) to find the important Zernike moments by employing the Genetic algorithm (GA) and (2) the classification of each character is performed using neural network. This way, the performance of the proposed technique is evaluated on two parameters (i.e., speed and recognition accuracy). Further, the efficacy of GA for selecting the moment features is assessed, and the efficacy of selected Zernike complex moments using GA is analyzed for handwritten Hindi characters. Here, the au...
2018 International Conference on System Modeling & Advancement in Research Trends (SMART), 2018
Developing an efficient character recognition system is supposed to be a very challenging researc... more Developing an efficient character recognition system is supposed to be a very challenging research problem. In the present work, an offline handwritten Hindi character recognition technique is proposed using Zernike moments as the descriptor of character image with a feature selection algorithm. For feature selection, use of the Genetic algorithm is proposed to reduce the length of the feature vector. The core idea of the paper is to first generate the significant Zernike complex moments and then to select the most relevant moments using the Genetic algorithm which are in turn used to classify the individual characters. The significance of low-order as well as high-order Zernike moments is also studied in recognizing the first ten consonants of Hindi script. Two resilient backpropagation classifiers are trained one for the feature vector without selection and another one for feature vector obtained after selection. The average character recognition accuracies obtained are 90% and 94...
2017 3rd International Conference on Advances in Computing,Communication & Automation (ICACCA) (Fall), 2017
The problem of character recognition may be categorized into two categories-recognition of printe... more The problem of character recognition may be categorized into two categories-recognition of printed characters and recognition of handwritten characters. Further, the problem of handwritten character recognition is categorized into two categories-recognition of offline handwritten characters and recognition of online characters. This paper presents an extensive overview of research work, which have been done for recognizing the offline handwritten characters using the various approaches like-ANNs, Fuzzy Logic, Genetic algorithms, SVM, KNN Hidden Markov Model (HMM), Bacterial Foraging, Clonal Selection Algorithm etc. and discuss their performance in terms of accuracy of recognizing the characters.
Handwritten signature verification is a very challenging and critical task. This work aims at pro... more Handwritten signature verification is a very challenging and critical task. This work aims at proposing an efficient offline handwritten signature verification model using writer independent approach. The prime focus of this work is on reducing the false acceptance rate of genuine signatures of writers while letting false rejection rate at a satisfactory level through ensemble of classifiers. The k-fold cross validation technique is used to develop ensemble of classifiers. The performance of the ensemble of classifiers, the support vector machine with polynomial kernel, is analyzed using the signature database of writers. The efficacy of geometric and uniform rotation invariant local binary pattern features is investigated to build a reliable writer independent offline handwritten system. The experiments exhibit 0.00 %, 0.00 % and 1.00 % false acceptance rate for random, simple and skilled forgeries, respectively while allowing false rejection rate 5.00 %.
Hyperspectral Image Processing a promising technique in remote sensing image analysis provides mo... more Hyperspectral Image Processing a promising technique in remote sensing image analysis provides more complete and detailed spectral information about land cover area and has a great potential to discriminate specific plant species using hundred numbers of contiguous bands. This study produces digital spectral library of Rice (Oryza sativa L.) varieties – Ratan (IET-1411), CSR- 10(IET-10349/10694), Haryana Basmati-1 (IET-10367), HKR-126 and CSR-13(IET-10348). The SAM (Spectral Angle Mapper), supervised classifier is used to develop the classified image at pixel scale, which discriminates rice varieties among 16 land cover classes with 86.96% classification accuracy. After pre-processing, the classification of rice crop at pixel scale across 155 calibrated spectral bands has shown promising result with 89.33% overall classification accuracy.
An efficient multiple classifier system for writer independent offline handwritten signature veri... more An efficient multiple classifier system for writer independent offline handwritten signature verification is proposed. Local oriented statistical information booster and histogram of oriented gradients feature descriptors are extracted from the signature image and a genetic algorithm is used to reduce the dimension of feature descriptors. To create multiple classifier system, two scenarios are generated. In scenario I, training set is divided into subsets using k-fold cross validation method and these training subsets are used to train the classifiers of multiple classifier system using same training algorithm for all classifiers. In scenario II, the classifiers of multiple classifier system are trained with multiple training algorithms. Unskilled and skilled forgeries are used to test the performance of multiple classifier systems. The performance analysis is carried out using support vector machine with polynomial kernel, quadratic kernel and gaussian radial basis function kernel....
Various offline handwritten signature verification systems using writer independent approach are ... more Various offline handwritten signature verification systems using writer independent approach are proposed by the researchers in last few years using numerous perspectives, like feature extraction techniques, feature selection techniques, classifiers used to develop the system etc. Despite the progressions in this framework, building classifier that can isolate the genuine and skilled forgery signatures is still a tough task. In this work, multiple classifier system is proposed to develop the writer independent offline handwritten signature verification system. To train the classifiers of multiple classifier system, feature vectors of the training set are partitioned into subsets and classifiers are trained using these subsets to preserve the diversity. The pixels lying on the elliptical curve paths are used to extract the features from genuine and forgery signature images. Two scenarios are proposed for the performance analysis. In the first scenario, the classifiers are trained usi...
Handwritten signature verification systems are studied using On-line and Off-line approaches. Due... more Handwritten signature verification systems are studied using On-line and Off-line approaches. Due to the lack of dynamic information, Off-line signature verification systems are more difficult to compare to On-line verification systems. There are two different approaches to deal with the problem of off-line signature verification. The first one is the writer dependent model (also known as personal model) and other is writer independent model (also known as global model). This paper presents a survey of off-line handwritten signature verification systems using writer dependent (WD) and writer independent (WI) approaches. Many recently used approaches of verification systems have been investigated. A conclusion of all these works and an outlook into the future work is also presented in this paper.
In the present study Satellite Image Processing (SIP) technique is applied on ASTER (Advance Spac... more In the present study Satellite Image Processing (SIP) technique is applied on ASTER (Advance Spaceborne Thermal Emission and Reflection Radiometer) satellite image. A comprehensive spectral library of rice crop varieties: Hybrid-6129 (IET 18815), Pant Dhan-19 (IET 17544), Pusa Basmati-1 (IET-18990) and Pant Dhan-18 (IET-17920) has been developed with Blue (0.56 nm), Red (0.66 nm) and NIR (0.81 nm) spectral bands. The conventional ASTER image is classified using ML (Maximum Likelihood) classifier. The PCA (Principal Component Analysis) transformation is also applied for feature extraction to select an optimum subset of data in term of classification accuracy. Four PCs (Principal Components) images selected for PCA classification. The conventional spectral classification accuracy for rice mapping is 79.5%, which is improved up to 84.5% with PCA classification.
The signature is a very significant trait of an individual which serves not only for the identifi... more The signature is a very significant trait of an individual which serves not only for the identification of an individual but also for establishing the genuineness of official documents. The aim of this work is to investigate the scope of geometric features to develop a proficient offline signature verification system through multiple classifiers using writer-independent approach. The key focus of this work is to minimize the false acceptance rate for the simulated forgery. The classification task is accomplished through Support vector machine with Gaussian radial basis function and polynomial kernel. The k-fold cross validation method is utilized to construct the multiple classifier system of diverse classifiers. The genuine and random forgery signatures are considered to train the classifiers of multiple classifier system whereas genuine, random forgery and simulated forgery signatures are utilized to perform the testing process. A public signature database named GPDS is utilized t...
2017 3rd International Conference on Advances in Computing,Communication & Automation (ICACCA) (Fall)
In this work, a writer independent offline handwritten signature verification model, also known a... more In this work, a writer independent offline handwritten signature verification model, also known as global model, for signature verification is proposed. Three classifiers, two back propagation Artificial Neural Networks and a Support Vector Machine with polynomial kernel are probed to develop the global model. Two databases of signatures from different writers are used to evaluate the performance of these classifiers in terms of false acceptance rate and false rejection rate. To develop the system geometric features and local binary pattern features are investigated. It is found in the study that Support Vector Machine outperforms the Artificial Neural Networks in developing handwritten signature verification system using geometric features and local binary pattern features.
International Conference on System Modeling & Advancement in Research Trends, 2019
Segmentation of words into isolated characters is the essential component in handwritten characte... more Segmentation of words into isolated characters is the essential component in handwritten character recognition systems. In this paper, the segmentation of Gurmukhi handwritten words into characters is presented. For this, horizontal and vertical projection features have been used to segment the characters from words. Simple words without upper and lower modifier of Gurmukhi handwritten text having three and four characters are considered in the present work. An overall accuracy of 91.4% on a dataset of 550 handwritten Gurmukhi words has been achieved in this work
International Journal of Computer Sciences and Engineering
Recognizing offline handwritten characters is a challenging problem and is considered to be more ... more Recognizing offline handwritten characters is a challenging problem and is considered to be more significant than the recognition of on-line handwritten characters. This study is undertaken to resolve the issue of offline handwritten character recognition for Gurmukhi script, one of the prominent scripts in the northern part of India. The Gurmukhi character images are presented using a single layer as well as multi-layer histogram of gradient features. Once the character images are represented using these features, one-against-all classification strategy, implemented through Support Vector Machine and k-Nearest Neighbour classifiers, is employed to recognize these characters. A dataset of 3500 handwritten Gurmukhi characters written by different writers is created and the scope of the Histogram Oriented Gradient (HOG) and Pyramid Histogram Oriented Gradient (PHOG) features is explored for the recognition of offline handwritten Gurmukhi characters. The simulation study reveals character recognition accuracy of 99.1% with SVM classifier for PHOG feature. The technique is robust to inter-class and intra-class variations present in the Gurmukhi script and has significant scope of application to the recognition of other scripts too
International Journal of Computer Science Issues
Offline handwritten character recognition is a very challenging area of research as handwriting o... more Offline handwritten character recognition is a very challenging area of research as handwriting of two persons may bear resemblance whereas handwriting of an individual may vary at different times. The character recognition accuracy depends on the ways the features are extracted from the samples and utilized to formulate the feature vector. In this paper, a novel technique 'TARANG' for feature extraction is proposed to recognize offline handwritten Hindi 'SWARs' (vowels). This technique for extracting features from an image is inspired by the natural movement of wave in a medium. A feature vector obtained by using proposed technique is used for the training of Backpropagation Neural Network and recognition rate as high as 96.2% is achieved.
International Journal of Systems, Control and Communications, 2018
Increasing traffic in mobile ad hoc networks (MANETs) demands high bandwidth. Mobility of communi... more Increasing traffic in mobile ad hoc networks (MANETs) demands high bandwidth. Mobility of communicating nodes is another crucial concern for such type of networks. Providing successful end to end communication in mobile ad hoc networks is a challenging task. In this paper, we aim at providing a bandwidth aware multipath reactive routing protocol for mobile ad hoc networks. The proposed protocol is an extension of ad hoc on demand multipath distance vector (AOMDV) routing protocol for mobile ad hoc networks. The proposed protocol, named as BAMR, attempts to discover paths with sufficient bandwidth and less link failures. The performance comparison of BAMR routing protocol is done with AOMDV and BAOMDV, an another bandwidth aware on demand multipath distance vector routing protocol for MANETs. Simulation results exhibit that proposed protocol BAMR significantly improves the performance of other two routing protocols and can be used more effectively for data transmissions in MANETs.
International Journal of Computer Sciences and Engineering
Indian Journal of Science and Technology
Int. J. Comput. Vis. Image Process., 2021
A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike c... more A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike complex moments like a tool to describe the characteristics of a character. Further, an algorithm for selecting the features is employed to identify the substantial image moments from the extracted moments, as the extracted moments may have some insignificant ones. Insignificant moments can increase the computational time and can also degrade the classification accuracy. Thus, the objectives of the study are twofold: (1) to find the important Zernike moments by employing the Genetic algorithm (GA) and (2) the classification of each character is performed using neural network. This way, the performance of the proposed technique is evaluated on two parameters (i.e., speed and recognition accuracy). Further, the efficacy of GA for selecting the moment features is assessed, and the efficacy of selected Zernike complex moments using GA is analyzed for handwritten Hindi characters. Here, the au...
Offline handwritten signature verification is a very challenging area of research as the handwrit... more Offline handwritten signature verification is a very challenging area of research as the handwriting of two people may bear similarity whereas handwriting of a person may vary at different times. The accuracy of handwritten signature verification system depends on the classifier system and the way of feature extraction. Keeping this point of view, four types of hybrid feature sets and three types of classifiers specifically support vector machine with polynomial kernel, support vector machine with quadratic kernel and decision tree are investigated for writer-independent offline handwritten signature verification in the present work. To obtain hybrid feature sets, local oriented statistical information booster, discrete wavelet transform, and histogram of oriented gradient feature descriptors are extracted and are coupled with each other. To create multiple classifier system, the training set is partitioned into subsets and these training subsets are used to train the classifiers of...
International Journal of Tomography and Simulation, 2019
Feature selection mechanism plays a vital role in attaining higher recognition rate in machine ba... more Feature selection mechanism plays a vital role in attaining higher recognition rate in machine based character recognition systems. Zernike moments and Zernike complex moments are found to be very beneficial as a basic tool to generate shape descriptors of a digital image, as they are capable of representing image features efficiently. These moments have been successfully utilized to solve some of the image processing problems and possess the property of rotation invariance which is a desirable aspect in many applications requiring a proficient offline character recognition system. This paper is devoted for inspecting the efficacy of Zernike moments and Zernike complex moments for offline Hindi numerals recognition. Feature vectors consisting of Zernike moments and Zernike complex moments are supplied to the resilient backpropagation neural network for training and different strategies are considered for the performance evaluation of these moments in the study. Character recognition...
Int. J. Comput. Vis. Image Process., 2021
A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike c... more A technique to recognize off-line handwritten Hindi character is suggested by employing Zernike complex moments like a tool to describe the characteristics of a character. Further, an algorithm for selecting the features is employed to identify the substantial image moments from the extracted moments, as the extracted moments may have some insignificant ones. Insignificant moments can increase the computational time and can also degrade the classification accuracy. Thus, the objectives of the study are twofold: (1) to find the important Zernike moments by employing the Genetic algorithm (GA) and (2) the classification of each character is performed using neural network. This way, the performance of the proposed technique is evaluated on two parameters (i.e., speed and recognition accuracy). Further, the efficacy of GA for selecting the moment features is assessed, and the efficacy of selected Zernike complex moments using GA is analyzed for handwritten Hindi characters. Here, the au...
2018 International Conference on System Modeling & Advancement in Research Trends (SMART), 2018
Developing an efficient character recognition system is supposed to be a very challenging researc... more Developing an efficient character recognition system is supposed to be a very challenging research problem. In the present work, an offline handwritten Hindi character recognition technique is proposed using Zernike moments as the descriptor of character image with a feature selection algorithm. For feature selection, use of the Genetic algorithm is proposed to reduce the length of the feature vector. The core idea of the paper is to first generate the significant Zernike complex moments and then to select the most relevant moments using the Genetic algorithm which are in turn used to classify the individual characters. The significance of low-order as well as high-order Zernike moments is also studied in recognizing the first ten consonants of Hindi script. Two resilient backpropagation classifiers are trained one for the feature vector without selection and another one for feature vector obtained after selection. The average character recognition accuracies obtained are 90% and 94...
2017 3rd International Conference on Advances in Computing,Communication & Automation (ICACCA) (Fall), 2017
The problem of character recognition may be categorized into two categories-recognition of printe... more The problem of character recognition may be categorized into two categories-recognition of printed characters and recognition of handwritten characters. Further, the problem of handwritten character recognition is categorized into two categories-recognition of offline handwritten characters and recognition of online characters. This paper presents an extensive overview of research work, which have been done for recognizing the offline handwritten characters using the various approaches like-ANNs, Fuzzy Logic, Genetic algorithms, SVM, KNN Hidden Markov Model (HMM), Bacterial Foraging, Clonal Selection Algorithm etc. and discuss their performance in terms of accuracy of recognizing the characters.
Handwritten signature verification is a very challenging and critical task. This work aims at pro... more Handwritten signature verification is a very challenging and critical task. This work aims at proposing an efficient offline handwritten signature verification model using writer independent approach. The prime focus of this work is on reducing the false acceptance rate of genuine signatures of writers while letting false rejection rate at a satisfactory level through ensemble of classifiers. The k-fold cross validation technique is used to develop ensemble of classifiers. The performance of the ensemble of classifiers, the support vector machine with polynomial kernel, is analyzed using the signature database of writers. The efficacy of geometric and uniform rotation invariant local binary pattern features is investigated to build a reliable writer independent offline handwritten system. The experiments exhibit 0.00 %, 0.00 % and 1.00 % false acceptance rate for random, simple and skilled forgeries, respectively while allowing false rejection rate 5.00 %.
Hyperspectral Image Processing a promising technique in remote sensing image analysis provides mo... more Hyperspectral Image Processing a promising technique in remote sensing image analysis provides more complete and detailed spectral information about land cover area and has a great potential to discriminate specific plant species using hundred numbers of contiguous bands. This study produces digital spectral library of Rice (Oryza sativa L.) varieties – Ratan (IET-1411), CSR- 10(IET-10349/10694), Haryana Basmati-1 (IET-10367), HKR-126 and CSR-13(IET-10348). The SAM (Spectral Angle Mapper), supervised classifier is used to develop the classified image at pixel scale, which discriminates rice varieties among 16 land cover classes with 86.96% classification accuracy. After pre-processing, the classification of rice crop at pixel scale across 155 calibrated spectral bands has shown promising result with 89.33% overall classification accuracy.
An efficient multiple classifier system for writer independent offline handwritten signature veri... more An efficient multiple classifier system for writer independent offline handwritten signature verification is proposed. Local oriented statistical information booster and histogram of oriented gradients feature descriptors are extracted from the signature image and a genetic algorithm is used to reduce the dimension of feature descriptors. To create multiple classifier system, two scenarios are generated. In scenario I, training set is divided into subsets using k-fold cross validation method and these training subsets are used to train the classifiers of multiple classifier system using same training algorithm for all classifiers. In scenario II, the classifiers of multiple classifier system are trained with multiple training algorithms. Unskilled and skilled forgeries are used to test the performance of multiple classifier systems. The performance analysis is carried out using support vector machine with polynomial kernel, quadratic kernel and gaussian radial basis function kernel....
Various offline handwritten signature verification systems using writer independent approach are ... more Various offline handwritten signature verification systems using writer independent approach are proposed by the researchers in last few years using numerous perspectives, like feature extraction techniques, feature selection techniques, classifiers used to develop the system etc. Despite the progressions in this framework, building classifier that can isolate the genuine and skilled forgery signatures is still a tough task. In this work, multiple classifier system is proposed to develop the writer independent offline handwritten signature verification system. To train the classifiers of multiple classifier system, feature vectors of the training set are partitioned into subsets and classifiers are trained using these subsets to preserve the diversity. The pixels lying on the elliptical curve paths are used to extract the features from genuine and forgery signature images. Two scenarios are proposed for the performance analysis. In the first scenario, the classifiers are trained usi...
Handwritten signature verification systems are studied using On-line and Off-line approaches. Due... more Handwritten signature verification systems are studied using On-line and Off-line approaches. Due to the lack of dynamic information, Off-line signature verification systems are more difficult to compare to On-line verification systems. There are two different approaches to deal with the problem of off-line signature verification. The first one is the writer dependent model (also known as personal model) and other is writer independent model (also known as global model). This paper presents a survey of off-line handwritten signature verification systems using writer dependent (WD) and writer independent (WI) approaches. Many recently used approaches of verification systems have been investigated. A conclusion of all these works and an outlook into the future work is also presented in this paper.
In the present study Satellite Image Processing (SIP) technique is applied on ASTER (Advance Spac... more In the present study Satellite Image Processing (SIP) technique is applied on ASTER (Advance Spaceborne Thermal Emission and Reflection Radiometer) satellite image. A comprehensive spectral library of rice crop varieties: Hybrid-6129 (IET 18815), Pant Dhan-19 (IET 17544), Pusa Basmati-1 (IET-18990) and Pant Dhan-18 (IET-17920) has been developed with Blue (0.56 nm), Red (0.66 nm) and NIR (0.81 nm) spectral bands. The conventional ASTER image is classified using ML (Maximum Likelihood) classifier. The PCA (Principal Component Analysis) transformation is also applied for feature extraction to select an optimum subset of data in term of classification accuracy. Four PCs (Principal Components) images selected for PCA classification. The conventional spectral classification accuracy for rice mapping is 79.5%, which is improved up to 84.5% with PCA classification.
The signature is a very significant trait of an individual which serves not only for the identifi... more The signature is a very significant trait of an individual which serves not only for the identification of an individual but also for establishing the genuineness of official documents. The aim of this work is to investigate the scope of geometric features to develop a proficient offline signature verification system through multiple classifiers using writer-independent approach. The key focus of this work is to minimize the false acceptance rate for the simulated forgery. The classification task is accomplished through Support vector machine with Gaussian radial basis function and polynomial kernel. The k-fold cross validation method is utilized to construct the multiple classifier system of diverse classifiers. The genuine and random forgery signatures are considered to train the classifiers of multiple classifier system whereas genuine, random forgery and simulated forgery signatures are utilized to perform the testing process. A public signature database named GPDS is utilized t...
2017 3rd International Conference on Advances in Computing,Communication & Automation (ICACCA) (Fall)
In this work, a writer independent offline handwritten signature verification model, also known a... more In this work, a writer independent offline handwritten signature verification model, also known as global model, for signature verification is proposed. Three classifiers, two back propagation Artificial Neural Networks and a Support Vector Machine with polynomial kernel are probed to develop the global model. Two databases of signatures from different writers are used to evaluate the performance of these classifiers in terms of false acceptance rate and false rejection rate. To develop the system geometric features and local binary pattern features are investigated. It is found in the study that Support Vector Machine outperforms the Artificial Neural Networks in developing handwritten signature verification system using geometric features and local binary pattern features.
International Conference on System Modeling & Advancement in Research Trends, 2019
Segmentation of words into isolated characters is the essential component in handwritten characte... more Segmentation of words into isolated characters is the essential component in handwritten character recognition systems. In this paper, the segmentation of Gurmukhi handwritten words into characters is presented. For this, horizontal and vertical projection features have been used to segment the characters from words. Simple words without upper and lower modifier of Gurmukhi handwritten text having three and four characters are considered in the present work. An overall accuracy of 91.4% on a dataset of 550 handwritten Gurmukhi words has been achieved in this work
International Journal of Computer Sciences and Engineering
Recognizing offline handwritten characters is a challenging problem and is considered to be more ... more Recognizing offline handwritten characters is a challenging problem and is considered to be more significant than the recognition of on-line handwritten characters. This study is undertaken to resolve the issue of offline handwritten character recognition for Gurmukhi script, one of the prominent scripts in the northern part of India. The Gurmukhi character images are presented using a single layer as well as multi-layer histogram of gradient features. Once the character images are represented using these features, one-against-all classification strategy, implemented through Support Vector Machine and k-Nearest Neighbour classifiers, is employed to recognize these characters. A dataset of 3500 handwritten Gurmukhi characters written by different writers is created and the scope of the Histogram Oriented Gradient (HOG) and Pyramid Histogram Oriented Gradient (PHOG) features is explored for the recognition of offline handwritten Gurmukhi characters. The simulation study reveals character recognition accuracy of 99.1% with SVM classifier for PHOG feature. The technique is robust to inter-class and intra-class variations present in the Gurmukhi script and has significant scope of application to the recognition of other scripts too
International Journal of Computer Science Issues
Offline handwritten character recognition is a very challenging area of research as handwriting o... more Offline handwritten character recognition is a very challenging area of research as handwriting of two persons may bear resemblance whereas handwriting of an individual may vary at different times. The character recognition accuracy depends on the ways the features are extracted from the samples and utilized to formulate the feature vector. In this paper, a novel technique 'TARANG' for feature extraction is proposed to recognize offline handwritten Hindi 'SWARs' (vowels). This technique for extracting features from an image is inspired by the natural movement of wave in a medium. A feature vector obtained by using proposed technique is used for the training of Backpropagation Neural Network and recognition rate as high as 96.2% is achieved.
International Journal of Systems, Control and Communications, 2018
Increasing traffic in mobile ad hoc networks (MANETs) demands high bandwidth. Mobility of communi... more Increasing traffic in mobile ad hoc networks (MANETs) demands high bandwidth. Mobility of communicating nodes is another crucial concern for such type of networks. Providing successful end to end communication in mobile ad hoc networks is a challenging task. In this paper, we aim at providing a bandwidth aware multipath reactive routing protocol for mobile ad hoc networks. The proposed protocol is an extension of ad hoc on demand multipath distance vector (AOMDV) routing protocol for mobile ad hoc networks. The proposed protocol, named as BAMR, attempts to discover paths with sufficient bandwidth and less link failures. The performance comparison of BAMR routing protocol is done with AOMDV and BAOMDV, an another bandwidth aware on demand multipath distance vector routing protocol for MANETs. Simulation results exhibit that proposed protocol BAMR significantly improves the performance of other two routing protocols and can be used more effectively for data transmissions in MANETs.
International Journal of Computer Sciences and Engineering
Indian Journal of Science and Technology