Algorithm Of Video Object Tracking

Current video surveillance technology has been widely used in transportation, public security, national defense, protection of privacy and other fields. Video images can provide people with the most realistic, objective, and valid information, which can effectively serve different people. With economic development and progress in information technology, privacy protection in video surveillance is getting attention and being investgated. Video surveillance target’s privacy to be protected is usually in motion, and its trajectory is unpredictable behavior, requiring the use of technical means to accurately track the moving object.Then the tracking target information will be encrypted to ensure effective protection of the privacy of an object, when surveillance video leaked for some reasons, it will not affect the safety of the protected object.This paper focuses on the still camera captured video moving target tracking algorithm and proposes an improved Camshift video moving target tracking algorithm. The methods of moving target tracking in video surveillance are introduced and an improved Camshift tracking algorithm is proposed for video surveillance scene complexity, and target illumination change.First, video image color space is converted from RGB to HSV, and the H and S components are extracted as in the traditional Camshift tracking algorithm.Combining them with image edge gradient feature, a joint three-dimensional color histogram is established.Second, video objects may move out of the camera view, the Kalman motion prediction module are employed to effectively improve robustness of the video surveillance moving target tracking algorithm in complex background.Experimental results show that using the improved Camshift video moving target tracking algorithm, moving targets under different light conditions can beaccurately tracked in real-time, even when the moving object and the background have similar colors or is partially occluded.