Vision Based Real-time Recognition of Hand Gestures for Indian Sign Language using Histogram of Oriented Gradients Features
Pradip Patel, Narendra M Patel · INTERNATIONAL JOURNAL OF NEXT-GENERATION COMPUTING · 2019
A sign language is the method of communication used by the deaf people where gestures are used to express meaning. Due to illiteracy of sign language by normal people, there exists communication gap between normal people and deaf people. Very little work has been done for recognition of Indian Sign Language due to lack of standardization and complexity of hand gestures. This resulted in need of automatic system that can recognize Indian Sign Language. Such system, developed via use of image processing and computer vision techniques, will help deaf peoples to communicate with normal people thus filling the communication gap. This paper presents vision based system for real time recognition of hand gesture for Indian Sign Language. The developed system is first trained using training data set. For this, from all the images of dataset, hand region is cropped by performing segmentation using thresholding in YCbCr color space. Histogram of Oriented Gradients (HOG) features of these cropped images are then computed and used to train the classifier. During testing, features of hand region of frames from real time video are presented to classifier for classification. Recognition rates of different classifiers like Support Vector Machine (SVM), K-Nearest Neighbors (KNN) and Linear Discriminant Analysis (LDA) are discussed.