Research on head counting method based on improved hough transform and multi-feature
SU Yan-man · Journal of Hebei University of Technology · 2013
It is very important for bus company's operations and scheduling to get the transit vehicle passenger information. Therefore, a new video-based traffic statistics algorithm with multi-information is proposed. Firstly, the target is extracted by the background method. Secondly, the transform passenger head is detected rapidly with the improved Hough algorithm, and the target locating is realized by the class round feature of head contour and the feature of head color.Finally, using the outline feature and color feature to identify head, regarding the circle of head as the moving targets and using the continuity of moving targets to do the head match and track and then implementing the accurate statistics. Large numbers of experiment results show that this method can accurate count the continuous passenger, have strong timeliness and can rapidly count the number of people passing the target area.