Hand Gesture Recognition Based on Convex Defect Detection

Human centered human-computer interaction technology tends to replace the traditional computer centered technology with its growing applicability to a wide variety of applications. Acting as a way of most natural communication between human and machine, vision based hand gesture recognition is becoming the pursuit of human-computer interaction. General vision based hand gesture recognition generally consists of sample capturing, image reprocessing, feature extraction and classification. Among these procedures feature extraction aims to detect and extract features that can be used to determine the meaning of a given hand gesture. The extracted features should be able to describe gesture uniquely and be robust to the shift and rotation of hand gesture in order to achieve a reliable recognition. In this paper, we propose a method to extract a series of features based on convex defect detection, taking advantage of the close relationship of convex defect and fingertips. This method is simple, efficient and free from gesture direction and position.