A Dynamic Gesture Prediction System Based on the CLCS Feature Extraction

Abstract

Real-time recognition of dynamic gestures is a problem for most of the applications nowadays. The prediction approach can be used as a solution for this. This approach uses an incomplete gesture input and it tries to predict which gesture the given input represents. This paper presents the application of the dynamic gesture feature extraction technique called Convexity Local Contour Sequence (CLCS) as the extractor for the prediction task. Two predictor systems are used to achieve this task and results are compared and discussed in this paper.