Part-by-part Human Movement Recognition Method Based on Non-negative Matrix Factorization

Abstract

The invention discloses a motion identification method for a human body by parts based on non-negative matrix factorization, mainly solving the problems in the prior art that feature extraction is complicated, the representational capacity is weak and the calculation amount is great. The motion identification method is realized by the following steps: (1) selecting a sample video I from a training video set X and detecting a motion interest point of each part of the human body and motion features of the corresponding part; (2) carrying out the non-negative matrix factorization on the motion features of each part of the human body; (3) carrying out cascading on the decomposed motion features of each part of the human body; and (4) carrying out the feature extraction on all the videos in the training video set X and a testing video set T to obtain a training sample feature set X' and a testing sample feature set T', and carrying out leaning training to obtain a classified result. The motion identification method can be used for accurately identifying human motions and carrying out video processing of video monitoring, human posture estimation and motion identification.

Description

technical field [0001] The invention belongs to the technical field of video image processing, in particular to a human body motion recognition method, which can be used for video monitoring and human body posture estimation. Background technique [0002] In recent years, human motion recognition has attracted people's attention as a major hot spot in the field of computer vision. It has wide applications and development prospects in video surveillance systems, driver assistance systems, and human-computer interaction systems. However, since the human body is non-rigid, factors such as its variability and diversity, clothing texture, lighting conditions, and self-occlusion seriously affect the effect of human motion recognition. Moreover, the resolution of video images is relatively low, and it is difficult to have obvious features. extract. Therefore, how to find the essential features of correctly expressing motion information from complex human motion and how to accurate...

AI Technical Summary

Problems solved by technology

[0004] At present, there are many methods of feature extraction: such as gradient histogram operator method HOG, scale invariant feature transformation operator method SIFT, accelerated robust feature operator method SURF, etc. The above feature extraction methods are suitable for two-dimensional images; The feature extraction methods in the image include: the gradient histogram extended to the three-dimensional space operator method HOG3D, the space-time block operator method Cuboids, the corner point detection extended to the three-dimensional space operator method Harris3D, etc., but the feature matrix extracted by the above methods The dimensions are relatively high, and the amount of calculation is large. In scientific literature, there are also many methods of using decomposition matrices to achieve dimensionality reduction, such as principal component analysis method PCA, independent component analysis method ICA, singular value analysis method SVD, vector quantization method VQ, etc. , these methods make the decomposed matrix values ​​have positive and negative values, which loses practical significance for practical problems, and the feature matrix used for dimensionality reduction is based on the whole, and there are certain information omissions in representing local information. thus reducing the recognition rate

Images