Real-time image sequences target tracking algorithm
|School||Xi'an University of Electronic Science and Technology|
|Course||Communication and Information System|
|Keywords||Tracking Related match Kalman prediction Mean-Shift PCA|
Real-time image sequence tracking (Real Time Object Tracking in Image Sequences), is a classic problem in computer vision , it refers to a group of real-time image sequences, depending on the desired target model , real-time image of the target to determine the location of the process. It was originally attracted the military's attention gradually being used in television -guided bombs , fire control systems and other military equipment . In recent years, computer vision and artificial intelligence, in-depth development in intelligent robots , electronic monitoring systems, traffic control and medical equipment , and other fields to be more widely used. This paper focuses on three target tracking algorithm, to find new ideas to solve the problems in the current algorithm for the future lay the foundation for further study . This thesis on a single image target tracking problem, a gray correlation matching template tracking algorithm based on Mean-Shift and kernel density estimation and tracking algorithm to analyze the relevant size of adaptive template matching algorithm problem and propose a region-based growth solution, in experiments have achieved good results, and the introduction of a noise variance adaptive Kalman algorithm , and improve the tracking accuracy ; on Mean-Shift algorithm in the kernel window width adaptive method was improved , and gives a way to improve the surface similarity kernel function ; final target and background intensity distribution similar to the problem of lost cause target was proposed based on PCA and edge detection target structure tracking algorithm, experiments show that the structure of a large tracking algorithm extent, improved the tracking number of frames , which has strong anti-noise performance.