The Use Of Machine Vision In Counting Of The People In Building Lift

In view of the current elevator overload detection to artificial, there is human factor interference. A statistical method based on machine vision for real-time detection of the number of elevators is proposed.The main research content of the real time count of the elevator is moving target detection, head area detection, target tracking and counting of 3 aspects. In motion detection, the background of the classical Vibe and the background of the update process are improved. According to the application of the system, we need to use this algorithm to achieve the segmentation of moving objects in video images. In postprocessing, the use of smooth removal of the image of the burr and noise, to ensure the accuracy of the follow-up head region detection.In the head detection, according to the contour characteristics of human head, a head detection algorithm is proposed, which can detect and locate the head of the monitoring area. The algorithm is suitable for the situation of the population density is not very high.In target tracking and counting, taking the head as the tracking target, the center point of the head ellipse is selected as the tracking feature points, which reduces the tracking difficulty. The method of common moving target tracking is studied, and a new method based on Calman filter is proposed.At last, the research work of this paper is summarized, and some problems are pointed out, and the future research and implementation of the work is prospected.