Kind of person traffic statistics system automatically selects the method classifier

The present invention discloses a method of automatically selecting human traffic statistical classifier system, video surveillance and belongs to the field of pattern recognition technology. The present invention comprises: a plurality of previously set imaging angle, a classifier trained off-line, respectively; online learning time for the current input image, performs the following operation for each classification are: Pedestrian detection target goals, the current has been detected and some tracking targets associating unassociated for target tracking using tracking algorithm to track the added unassociated detection target tracking destination queue; destination queue tracking target qualified for counting; if all the counted classifier target the number did not exceed the threshold, the process continues to the next image; selecting target count up the flow of people classified as optimal statistical classifier used in the system. The present invention provides a method capable of human traffic statistics system automatically selects the optimal classifier, avoid manually set classifier inaccuracy and inconvenience issues.