The invention discloses an intelligent ridership automatic statistical method based on a self-adaptive threshold value. The method comprises the following steps that a video sequence of a detecting zone is collected in real time through monitoring equipment, and a foreground moving target is extracted from the video sequence obtained by collecting; according to a practical application scene and the distance between a camera and the detecting zone, the position and the size of a ridership detecting frame are set; based on video information and the ridership detecting frame, the self-adaptive threshold value is generated by unsupervised learning; a method based on the geometry principle is used for judging the behavior of passenger getting-on or getting-off on the ridership detecting frame; and according to the judging results and the self-adaptive threshold value, the ridership is subjected to judging and statistics. The method is simple and efficient, practicability and portability are high, and the method is suitable for intelligent automatic statistics of the passenger car ridership.