Human body target intrusion detection method

The invention relates to a human body target intrusion detection method. According to the method, firstly, a motion target of each frame of a video is detected by use of an improved background subtraction method; then a shape feature of a target is extracted, and a suspected human body target is screened out according to a human body shape feature; next, a motion feature of the target is extracted, and the suspected human body target is screened out according to a human body motion feature; and finally, an Haar feature of a suspected human body area is extracted, and whether a target area contains a human body is detected according to a human body space gray level distribution feature. The invention provides the human body target intrusion detection method based on multi-feature fusion, so that the human body detection speed is improved, the false alarm rate is lowered, whether the video has an intruding human body target can be detected and monitored intelligently, and the safety and intelligence level of an intelligent monitoring system can be improved.