Research On Human Gesture Detection And Recognition Methods

With the rapid development of science and technology,computer vision technology has become a very important and key technology in People's Daily life,students' learning and work in all walks of life.Among them,gesture recognition,as the most direct and natural technology of humancomputer interaction,is widely used in People's Daily life and has greatly improved people's life.This paper mainly studies human gesture recognition methods based on different feature fusion and image preprocessing to improve the recognition rate of gesture recognition algorithm.The main work of this paper are as follows:1.In the process of extracting gesture features,it is often affected by skin color,hand shape,lighting conditions,background differences and other factors,which makes it difficult to specify the appearance of hands by a single feature.Therefore,this paper proposes an enhanced fusion HOGLBP feature detection and SVM classification detection method to replace the uniform LBP feature with the improved mb-lbp.Experiments show that this method is more accurate in light condition,skin color and complex background.2.Kinect sensor device brings new experience to a large number of users.As the Kinect sensor has been upgraded,it has become more powerful.In this paper,Kinect sensor is used to obtain gesture depth image and color information,and skin color model is used for secondary segmentation to capture more accurate hand images.Then,HOG-MBLBP,a new fusion feature proposed in this paper,is used for feature extraction,and finally SVM is used for classification and recognition.Experiments show that the proposed method has higher recognition rate than other methods.3.As an important direction of human-computer interaction,dynamic gesture recognition has broad requirements in various fields.Compared with static gesture recognition,dynamic gesture change has more diversity and poor detection effect.Therefore,this paper proposes a gesture detection algorithm that combines the improved three-frame difference method and background difference method to detect the interference of skin color and brightness changes on dynamic gestures under the background of non-single factors.The edge detection is integrated into the three-frame difference method,and the obtained image has good contour information,making up for part of the "void" phenomenon.At the same time,the hybrid gaussian background difference algorithm is fused to supplement the improved three-frame difference method,so that the detection results have the characteristics of continuous contour and relatively complete internal information at the same time.Experiments show that the gesture detection algorithm proposed in this paper has high recognition rate and low false detection rate.