Research On Algorithm Of Infrared Small Target Detection And Tracking In Image Sequence

Infrared imaging technology has many advantages such as stronganti-interference ability, well-concealment and good weather adaptability, it gets awidely use in the military and civilian applications fields. Modern warfare putforward higher requirements to the weaponry and early warning systems. How toimprove the detection distance in order to earlier and faster find and track the targetis a hot and difficult problem in the field of infrared images processing currently.Because of the long imaging distance, the target forms a small area in the image. Bythe great impact of noise and background clutter, how to achieve in complexbackground conditions has important significance for the study of robust infraredsmall target detection and tracking. This dissertation focuses on infrared small targetdetection and tracking technologies in-depth. Following are the major wok andresults of research.In the image preprocessing, this paper studied the classical algorithms ofbackground suppression. To overcome the limitations of traditional bilateral filteringand windowed bilateral filtering algorithm, this paper presents the imagepreprocessing algorithm based on the combination of Tophat transformation andwindowed bilateral filtering. Experiments confirmed that this algorithm effectivelyimproves the system background suppression and target detection performance. In terms of target detection, the paper studied the classical thresholdsegmentation algorithm, scale space theory and DoG scale space algorithm. Thispaper completed two single-frame detection experiments using DoG algorithm, oneis the specified scale target detection，the other is non-specified scale targetdetection.Base on the DoG algorithm, we proposed a new multi-frame detectionalgorithm: the combination of DoG algorithms and pipeline filtering algorithm. It iscalled diameter adaptive pipeline filtering——PDAF algorithm.On target tracking, this paper got a detailed study of the Kalman filteralgorithm and particle filter algorithm. We propose a new algorithm that combinesthe advantages of Kalman filter and DoG algorithm. Using Kalman filter to implythe initial estimates of position of the moving target, and then use the DoG algorithmto get the accurate location and scale parameters of the target in the tracking wavegate. We use the obtained parameter values as the observed values to update theparameters of the Kalman filter, which is conducive for Kalman filter to accuratelypredict the location of small targets in the next frame. Experiments show that thealgorithm can effectively improve the positioning accuracy of the Kalman filter.There is a scale evolving issue in target tracking process.In order to give fullplay to the advantages of particle filter in tracking large targets, we design analgorithm switching strategy base on scale space theory: using the DoG algorithm toget the target scale, when the target scale is below the threshold using the Kalmanfilter algorithm for target tracking, when the target scale is above the threshold usingparticle filter algorithm. The proposed algorithm can effectively improve thetracking performance of the system.Finally, in order to meet the requirements of real-time target tracking, wedesigned a DSP+FPGA-based hardware platform tracker and validated algorithmsand hardware platform.