Gesture Recognition Method Based On Micro-motion Video Key Frame Extraction And Its Application In Acupuncture Recognition

For human motion recognition problems,the existing methods are mostly based on human limbs,hands,elbows,etc.For the recognition of finger movements and other micro-motions,The recognition effect obtained by common dynamic recognition methods is not ideal,and recognition often fails.Problems such as accuracy or inability to capture the movement of fingers.The motion information of micro-motions is more expressed in the motion state of the fingers.The motion state of themicro-motion gestures and the length of the motion cycle are used to determine a complete motion.Key frame extraction and feature processing for such repetitive micro-motion videos Provides a new direction.The key frame sequence is an important form of expressing video content.In recent years,common key frame extraction algorithms have encountered incomplete key frame extraction when solving videos with small motion range of target objects,repeated content,single shot and unchanged scene.Problems such as insufficient effective information extracted and redundancy of extracted frame images.Perceptual hash algorithm is a common method in the field of image search,but it provides a new solution to the problem of micro-motion key frame extraction in this paper.In the micro-motion video,the spatiotemporal features of the target object's movement contain all the information of its actions.The effective extraction of spatiotemporal features in the video can make the network training more effective.Convolutional neural networks are often used to extract spatial features of images,but in the face of video sequence feature extraction,they often cannot effectively process time series features,and the appearance of long and short-term memory networks provides time feature extraction for video sequences.New solutions.Therefore,this paper proposes a key frame extraction algorithm based on micro-motion gestures,which can improve the accuracy of key frame sequence extraction of micro-motion gestures.A key frame extraction algorithm based on perceptual hash algorithm and Hamming distance is constructed.A calculation method for judging the starting frame and the ending frame is given;combined with the neural network recognition method for micro-motion gestures,a hybrid network model is constructed.The proposed algorithm was tested using public data sets,and the experimental results verified the effectiveness and advantages of the algorithm.The key frame extraction algorithm and hybrid network model are used to process the local acupuncture manipulation data set,which verifies the feasibility of the algorithm and model in the recognition of acupuncture manipulation.This paper presents an effective method for micro-motion gesture recognition,and also provides a new algorithm for video key frame extraction.Combined with the content of deep learning,a computer model for the recognition of acupuncture is established.This model is acupuncture.The scientific research and quantitative teaching provide new directions.