This paper focused on analyzing the algorithm of multi-object vehicles detection and picking up in the vision field
when the camera is fixed. From the perspective of given condition and environment, combining the key and difficult
points, according to the continual vision sequence recorded by camera, discussing and analyzing the several realization
algorithms of the multi-moving vehicles detection, extraction, identification and classification. On this basis we present
modification and improvement of the relevant algorithms. Therefore in the limit of given condition, effectively increasing
the real-time, object, robustness and adaptability multi-moving vehicles detection and extracting.