Performance Comparison of Transforms using Row Mean and Column Mean for Hand Gesture Recognition

Abstract

computer interaction is a major issue in research industry. In order to offer a way to enable untrained users to interact with computer more easily and efficiently gesture based interface has been paid more attention. This paper presents a new approach for hand gesture recognition. An approach consists of three modules: a) Preprocessing of the image b) Feature extraction c) Pattern matching for gesture recognition. Feature extraction is based on feature vector of transformed image using Discrete Cosine Transform, Walsh Transform, Haar transform and Kekre's transform. This transforms are applied on column mean and row mean of the images and various percentage of feature vectors are generated such as 100%, 50%, 25%, 12.5% and 6.25%. Results found to be better than existing system.