The study of vehicle existence detection based on inter-frame similarity evaluation algorithm

In order to detect vehicle effectively without affected by partial illumination and texture change and avoid image preprocessing, one vehicle detection method based on inter-frame similarity evaluation algorithm (IFSEA) is proposed in this paper. Based on quasi-shot segmentation theory, RGB space histogram intersection algorithm and YUV space brightness histogram evaluation algorithm are combined together to weight the size of the visual content change in IFSEA. The ultimate inter-frame similarity evaluation score is designed by the intersection of results got by IFSEA. So vehicle can be detected according to the score of evaluation criteria by signal bottom detection and threshold method. Experiment shows that not only the vehicle's existence can be detected accurately but also the process of vehicle's entry and exit can be revealed by IFSEA.