Application Research Of Adaptive Algorithm In Visual Target Tracking

Target tracking is by computer instead of the human eye machine vision recognition,tracking and measurement made on the target,and further do graphic image processing,and computer processing is more suitable for human observation or transfer to explore the process of the image detection instrument.The application of Mean for Shift tracking algorithm research as the main line of the transportation system,explore the tracking algorithm depends on the bandwidth adaptive mean shift algorithm and Calman filter depends on the target,target tracking algorithm and rigid verification activities focus on the complicated situation.The main obstacle is to track project exploring the problem of complicated traffic targets,with visual tracking method,combined with the following aspects:the characteristics of the target,like the choice of target tracking made deep exploration,for the complicated situation prompted the target tracking is robust and timely method demonstrated better.In this paper,we first analyze the mean shift target tracking method,and then put forward the method of defect rectification.First of all,the paper analyzes the basic principles and algorithms of the common target tracking technology.It is found that the mean shift algorithm is the address of tracking the target in the new area.If the target transfer rate is relatively rapid,for the new domain with the center of the target region in the center of the scope of the original far,existence interval of two continued across the bandwidth of the H window frame transfer if a goal from the center,will have occurred in the mean shift tracking algorithm with often resulting in invalid state tracking.In addition,the rapid transformation of the target size will also cause the target to be lost.The mean shift algorithm,in the aspects of carrying out the process of tracking,the tracking window size is basically fixed,not in accordance with the target size to make corresponding rectification,for example,a car from afar,the car body will gradually increase,but with the tracking window size is fixed,so in the tracking window we will only show the local auto target.With a part of the data to match the overall data,the results can only match the failure,so that the tracking target lost.Secondly,in order to track the link in the form of the target size of the continuous transformation of the status of the principle,which makes the bandwidth window adaptive rectification approach introduced.Bandwidth window its size by the calculation of the target range of the size of the screen,the size of the bandwidth window is the target area of the size of the change to carry out the immediate refresh.Finally,to understand the general situation in the mean shift algoritlhm can quickly correct tracking,especially for tracking general occasions,in occasion of the target relative effect is to track the target size did not change or set intervention is not much.There are over the situation(in the range of target image tracking process,there will be some different size,stable new gray range,let the tracked target range of image size will be smaller due to cover or short time invisible),loss of the tracking target situation is likely to occur.Robust performance to make the target tracking algorithm implemented in the cover condition,to cover the situation in the target tracking and therefore be presumed adopted the mean shift algorithm by means of adaptive Calman filter and modified fusion.