The Designing And Implementation Of Movement Target Tracking System

Moving target tracking has always been the difficulty and hotspot in the field of computer vision and its application field is very broad, such as in the field of video surveillance, human computer interaction field, and intelligent transportation field and so on. Under the circumstances such as in a variety of complex environment or the shadows shade, scenes changed, illumination changed, target transient disappeared in the field of vision and goals form changed, tracking result still can be very good, which is the problem that general vision research workers to solve currently and focus in application.Based on the research at home and abroad, aiming at problems of two mainly tracking aspects includes moving target detection and moving target tracking, this article’s mainly research contents and innovations are as follows:1、In the moving target detection aspect, comparatively analyzed three common method of detecting the moving target:optical flow method、inter frame difference method and subtraction method. And through the experiment, it was found that background subtraction method can obtain more complete target contour and faster detection because of its less computational complexity.2、In the moving target tracking aspect, this experiment use traditional tracking method—camshift tracking method, and result indicated that it has strong robustness、adaptability and anti-shielding ability. Aiming at semi-automatic tracking mode, I make some improvements that get centroid area histogram through the centroid of moving object to realize automatic tracking moving targets.3、To achieve better tracking performance, a new tracking method named TLD tracking was put forward. TLD algorithm is a kind of system that can detect, online learn and track at the same time. It was based on Pyramid LK optical flow and joined detection module, learning module and synthesis module. This method can effectively solve the problem of target occlusion, deformation and disappearance of. And the experimental results showed that this method can realize complex background or moving background for long time tracking of moving target, and had good stability and accuracy.