A Research On Target Recognition And Tracking Technology With FPGA

Computer vision technology which has been wildly used in military and civilian areas has been rapid developed in recent years.An important branch of computer vision is identification and tracking technology of targets.This paper mainly studied the method of moving objects' detection and tracking based on template match,which plays primary role in precise terminal guidance of missiles.First,the thesis gave the current research situation about detection and tracking technology of moving target,and introduced the image preprocessing technologies which are used in objects' identification and tracking.This paper also compared and analyzed their performance in different environment,while implemented these methods in FPGA.Top it off analyzed there method about target detection and their application condition,while comparing their detection results,resource consumption and detection time.Second,on the foundation of object identification algorithm,this paper focused on the tracking algorithm of moving target based on the template matching.First of all,this paper studied the classical fixed template matching algorithm and completed software simulation.At the same time,this paper analyzed the disadvantages of fixed template matching algorithm and factors which influence the tracking accuracy.Then the author proposed a tracking algorithm based on image hash and wavelet transform to solve the decline of tracking accuracy which is caused by the deformation and occlusion of the target,which means the image hash which is calculated through DCT transform,is treated as the basis for updating the template.Searching the best matching area in the low frequency sub-image which can obtain by wavelet transform reduced the calculation and improved the matching speed.Experiments show that this algorithm not only has a good performance when there is rotation,zoom and other deformation of target but also raise the system's real-time performance.Finally,this paper implemented target tracking algorithm in the Spartan-3E series of FPGA system,and experiment the tracking effects in single background,complex background and environment where contain interference objects.The results show that,the system can correctly track the moving target in the three conditions with a good real-time performance..