An analysis of edge detection as a feature extractor in a hand gesture recognition system based on nearest neighbor

Hand gestures, as a part of human body language, can be used for many purposes. By means of a hand gesture recognizer, we could communicate with machines using our hand gestures. A recognition system typically consists of pre-processing steps and a classifier. This paper presents an analysis of using edge detection and/or histogram equalization in the pre-processor by examining the overall performance of the hand gesture recognition system. Nearest neighbor classifier is used as a classifier in the recognition system. The system aims to classify the input images into one of six classes. Each class represents a different command to a machine. The hand gesture images are taken using a web camera under controlled condition and a uniform white background. The system performance is measured by using cross validation method. The experiment results show that using histogram equalization and edge detection as feature extractor lowered the average of accuracy.