Application of Multi-Camera Tracking and Positioning Technology in Substation
Guifeng Zhang, Zhiren Tian, Yongli Liao, Song Wang, Wutao Chen · 2019
With the development of camera hardware technology and computer vision algorithms, visual positioning technology has been paid more and more attention. Most of the research on visual positioning technology is based on single camera and single picture. that is, only the target tracking appears in the shooting range of a certain camera, the coverage and the obtained picture information are limited. The multi-camera-based visual positioning technology can not only provide a wider field of view but also overcome the problem of target occlusion by the cooperation among the cameras. The multi-camera visual positioning technology studied in this paper mainly applies several classical algorithms in this field. The KNN-based background segmentation algorithm is used to detect moving targets, and the KCF algorithm is used to achieve continuous and reliable tracking of targets. The SURF feature detection algorithm is adopted. Target matching is achieved by matching with FLANN, and finally the position and motion trajectory of the target in the bird’s eye view are obtained based on the planar homograph. The multi-camera visual positioning technology can be used to obtain the real-time position of the personnel in the substation, which can prevent the staff from entering the charging interval, and avoid the dangerous situation that the outside personnel enters the substation illegally.