A Camera PTZ Control Algorithm For Autonomous Mobile Inspection Robot

Abstract

Traditional inspection robots usually carry fixed visual inspection equipment. Due to the limited monitoring field of view, the target cannot be tracked and photographed. Therefore, an intelligent pan-tilt tracking and snapping system is designed in this paper. The PTZ camera is combined with the kernel correlation filter tracking algorithm. The position information and scale information of the target are calculated by the tracking algorithm, and then the rotation and magnification of the PTZ camera are adjusted in real time through the control program, so that the target is always in the center of the monitoring field. The PTZ control algorithm is divided into three parts: initialization part, pan / tile control part and zoom control part. Based on the traditional kernel correlation filter tracking algorithm, a multi-scale tracking method is added, which solves the adaptive problem of target scale change in the original algorithm. It can be well applied to the tracking scene where the target scale changes or the pan tilt camera rotates and zooms.