Research And Design Of Tracker Based On Improved CAMSHIFT Algorithm

Object tracking technology based on video was used in the military field firstly, in recent years with the high-speed development of computer technology, as well as evidently improvement in the capability of image capture hardware, the object tracking technology has been widely used in a variety of products for civilian use that more and more closely related to people’s daily lives.Camshift (Continuously Adaptive Meanshift) algorithm based on image color characteristics is a very wonderful object tracking algorithm, due to the non-sensitive for the object deformation and the real-time characteristics.By used improved Camshift algorithm as the theoretical basis and a tennis match as the main test scene, this article designed a method for tracking object movement trajectory, and thus draw the player’s activity area and the moving distance of object. Due that traditional Camshift algorithm has some shortcomings, this article gave some improved action. In the object initialization phase, the traditional Camshift algorithm just use the mouse to take a single rectangular box that contains the actual object as the tracking object,that don’t have a standard what’s space it should be include, and that owing to the irregular contour of the object, rectangle area inevitably mixed with some of the environmental background information that decline the accuracy of the object color histogram.This paper take a new method to solve this problem that selected internal-rectangle and external-rectangle by two times. It is able to decline the error caused by environmental background effectively. In the tracking stage, the pixels of the object area need to be filtered due to image quality issue, however, there isn’t a standard method to selected suitable threshold,that may lead the size of the object area disordered changed, especially for far distance object or the object be blocked partly. Such changes will often result in loss of the object, and can’t be restored once again. This paper proposed an adjustable size of the searching window that function according to an adaptive filter threshold method to protect the stability of tracking system that may affect by the disordered change of searching window. Verified by experimentation, compared to the traditional Camshift algorithm the improved algorithm significantly improved the stability and accuracy of object tracking.Tennis is very popular sport with the people, but how to evaluate players on the field performance, and how timely and accurate grasp of the player running distance, regional, and other information is basically relying on people’s subjective feeling so far, and there isn’t reliable data to do qualitative analysis. Similarly, in the game process, in order to timely acquire the opponent’s status and features as much as possible, the player need to learn some parameters such as moving distance and activity area to make further analysis.Based on the above problem, by used of improved Camshift algorithm, the subject designed a tracker for the tennis athletes to measure running distance and to visualize the player’s main running region on the image. The tracker finally verified in VC++software environment.