Study the improved CAMSHIFT algorithm to detect the moving object in fisheye image

摘要

Fisheye lens has been applied more and more in the digital video surveillance system because of large field of view, but the fisheye image is distortion greatly which restricted its further application. According to the characteristics distortion of circular fisheye image, this paper studied an improved Camshift object detection algorithm and applied to the moving object tracking and detection in the fisheye surveillance system. Firstly, the principle of the Meanshift algorithm was discussed, and then the Camshift algorithm which based on the Meanshift algorithm was studied. Lastly, the improved algorithm of Camshift was designed which based on the Camshift algorithm and the characteristics of the fisheye image, and validity of the algorithm by calculation. The experiment results shows that the improved Camshift algorithm can detecting and tracking the moving object in fisheye image stabilization and effective, whether it is in the middle position or on the edge with compression seriously. It has a very good application in the intelligent and nothing blind-zone surveillance system.