Real-Time Multi-Channel Video Stitching Based on Perspective Transformation in Distribution Cabinet Monitoring Scene

At present, the widely used video stitching technology based on image registration has many problems when dealing with distribution cabinet monitoring images, such as a small number of detected feature points, similar elements leading to feature point matching errors, and concentrated distribution of feature points leading to low accuracy. In order to solve these problems, based on the characteristics of distribution cabinet, this paper uses perspective transformation instead of image registration to complete image alignment, which effectively improves the problem of image distortion caused by low accuracy of image registration method in distribution cabinet video stitching scene. In addition, to improve the speed of preprocessing, offline calibration parameters method is adopted. In order to adapt to the uneven resolution in the actual industrial control environment, the steps of adjusting the screen resolution are integrated into the perspective transformation. Finally, through the comparison with Scale-Invariant Feature Transform (SIFT) algorithm and engineering application, this paper realizes the high quality real-time stitching of 7 cameras.