The study aims to improve the performance of the recognition algorithm for human behaviours. An improved Support Vector Machine (SVM) behaviour recognition method based on dynamic and static characteristics is studied, and video surveillance is used to track and test human targets. In video frames, the average background method is used to model the static background, and the optical flow is used to model the dynamic background. In terms of target tracking, a multi-feature particle filter is used. And an improved Fuzzy Support Vector Machine (FSVM) is used for behaviour recognition based on the combination of dynamic and static characteristics. The results show that the integration of dynamic and static characteristics of human behaviour can comprehensively show human behavioural characteristics. And experiments are carried out on the KTH data set, and the detection accuracy increases by 2.05%.
展开▼