Real Time Hand Gesture Recognition Using Histogram of Oriented Gradient with Support Vector Machine

Numerous applications use Hand gesture recognition (HGR) as an essential part of their system. One of the most important Human Computer Interaction (HCI) applications using HGR is to read and process sign language. There are various methods used for Hand Gesture Recognition. In this paper, we have used Kinect sensor to take real-time images of hand gestures. Then, we have applied Histogram of Oriented Gradients (HOG) on the images to extract hand features. A dataset of 3000 images from Kinect Sensor of 30 different people was created. Finally, we have applied Multi support vector machine to classify the hand gestures. This helped us to achieve an accuracy of 94.3% for real time hand gesture recognition in a cluttered environment.