Video traffic flow detection method by block background modeling

The invention discloses a video traffic flow detection method by block background modeling. The method is characterized in that: the method comprises the following steps: I. after determining the location of a virtual loop in a video image, dividing the virtual loop into three blocks by a video signal input module; II. performing motion frame difference detection on image data by a motion frame difference detection module and detecting the motion statuses of the three blocks of the virtual loop; and III. carrying out background subtraction detection on the image data by a background subtraction detection module, and updating the current background model according to the background subtraction detection result in combination with the motion time sequence status of the previous moment loop. In combination with the practical application scenes of the virtual loop and according to the motion status time sequence of continuous blocks, the method can rapidly learn the background, can rapidly switch to a new background under the condition of background change, and can be well adaptive to the background modeling requirement of the traffic flow video detection of the virtual loop.