Structural Classification Of Gesture Recognition Algorithm Research

Interaction between human and computer is playing a important role in our daily life. Especially in recent, the Human Computer Interaction tools is becoming hot with the development of computer technology, that make us use the nature input method instead of mechanical method such as voice and gesture, so voice and gesture recognition is becoming focus of our research.My work is part of Shanghai Nature Science Fund research task-Gesture Recognition& Combination and focus on gesture recognition. Gesture recognition involves pattern recognition, digital image process, recognition science. The whole recognition process can be separated into image preprocess, classification, feature extraction and recognition.In the part of preprocess phase, we should sharp the image and then use the maximum variance binalization, and after that is preliminary classification, eliminate noise, and the last is extract gesture contour.Extract feature based on structure should extract gesture property based on gesture contour and classify them and then extract property again, classify again, so it works in this way until we couldn't classify them. But Extract feature based on statistics use SVD method.In recognition process, I use structure classification method. If it doesn't work I choose statistics method in which a distance classifier has been built up for gesture recognition. Then the distance between the unrecognized vector and vectors in standard gesture library can be calculated for judgements. We classify the image into the sample class that has theshortest distance. Experiment has proved this method can got 84% accuracy.DutaoZhuDirected by: Linquan Wang...