Real-time multidirectional pedestrian counting and tracking method

Abstract

The invention discloses a real-time multidirectional pedestrian counting and tracking method. The method includes: acquiring depth images and color images by adopting a depth camera; extracting moving block masses from the depth images; extracting Harris angular points from the color images; calculating optical flow vector groups through the Harris angular points mutually matched in two frames of images which are mutually spaced by one frame of image; clustering the optical flow vector groups into a pedestrian location point; calculating optical flow vectors of the pedestrian location point. A pedestrian-crowded scene and a pedestrian sparse scene can be monitored and counted with the method which is high in practical application value, small in calculating amount and low in error recognition, the signal camera is used, convenience in installation is achieved, and impact on the monitoring environment is quite small.

Description

technical field [0001] The invention belongs to the field of computer vision, in particular to a real-time multi-directional pedestrian counting and tracking method. Background technique [0002] In the field of intelligent video surveillance technology, the research on pedestrian counting has always been the core of attention of those skilled in the art. This is because, by counting the pedestrians in the scene, when an emergency occurs, it can effectively disperse and evacuate the flow of people according to the distribution of the number of people in the scene, and minimize the harm caused by the emergency. [0003] At present, there are mainly two methods of pedestrian counting: [0004] 1) Tracking of moving blobs provided in OpenCV; [0005] The method mainly consists of four modules: foreground detection, blob detection, blob tracking, and trajectory processing. The moving clumps are obtained by using the moving target detection, and different tracking methods are ...

AI Technical Summary

Problems solved by technology

However, this method has its fatal shortcomings: a. It has a strong dependence on the establishment of the background model and the extraction of the foreground target. When the background in the scene changes rapidly, it cannot show good performance; b. If the moving target If the moving target stays in the background for a relatively long time, the moving target will be included in the background model; c. Especially when an algorithm with relatively high detection accuracy is selected, the real-time performance of the method cannot meet the requirements, let alone tracking and count up

However, in practical applications, the method of only using optical flow cannot overcome the influence of adverse conditions such as occlusion, multiple light sources, transparency, and noise. Therefore, the gray-scale conservation assumption of the basic equation of optical flow cannot be satisfied, and the correct The optical flow field, and the calculation method of optical flow is very complicated, a large number of calculations cannot meet the real-time requirements at all, and it is impossible to realize the real-time multi-directional pedestrian counting method

Images