Hand Gesture Recognition Based on MEB-SVM

In this paper, we propose a novel static hand gesture recognition method, which is based on a new support vector machine (abbreviated as SVM) classifier. SVM is a classification method based on statistics theory. Typical SVMs can be sufficient to deal with small scale data, but these methods cause a lot of computation in quadratic programming while dealing with non-linear problems. SVM combined with MEB (minimum enclosing ball) is a powerful tool. It reduces the massive computation and also can separate all kinds of vectors in a hyperspace efficiently. First and foremost, image segmentation must be done before hand gesture recognition. We adopt mean shift, which is using skin color for the image feature. Finally using MEB-SVM to classify gestures, and achieve the aim of recognition.