Research On Unmarked Hand Gesture Recognition Based On Computer Vision

With the rapid development of computer hardware and software technology,the relationship between people and computer is becoming more and more closely related,and the way of human-computer interaction is becoming more and more diverse.The unmarked gesture recognition system based on computer vision has been paid more and more attention because of its friendly and natural characteristics.However,there are still some difficulties,such as the poor recognition rate and the poor real-time performance.Based on the analysis and summary of the research status at home and abroad,this paper studies from three aspects:gesture detection,gesture tracking and gesture recognition,the main contents in the paper are as follows:1.Hand gesture detection.The first step of gesture according to the color feature,first studied the gesture segmentation method based on color ellipse model,respectively,by comparison and analysis of skin color clustering in RGB,HSV,YCbCr color space,and then combined with the characteristics of the connected domain based on modified ellipse model,can achieve complex background distinction between skin color and facial gestures the second step in the skin;skin color segmentation based on texture feature is studied according to the gesture,fast Adaboost classifier based on Haar-like features,using multi feature combination,effectively reduce the amount of data in image processing,and can improve the gesture recognition rate.2.Hand tracking.Study on the Mean-shift object tracking algorithm,and hand the zoom,perspective and other issues,put forward a combination of SIFT feature gesture tracking algorithm,the algorithm of spatial correction of the gesture region,can effectively improve the gesture tracking accuracy and has good robustness for the hand of the zoom and brightness change.3.Hand gesture recognition.A static gesture recognition algorithm for system instruction control is studied.The gradient feature of gesture is classified according to the gradient histogram feature of the gesture,and the gradient feature is classified by support vector machine,which can perform static gesture recognition.Then,the basic gesture library is built to train the classifier,and the gesture recognition in the opponent potential library is carried out by training classifier.The accuracy is 94.2%,and the recognition accuracy and speed are better than those in the literature.