Research On Infrared Small Target Detection And Tracking Technology

As the key technology of infrared guidance, search-track and early warning system, infraredsmall target detection and tracking technology has been a hot research topic in the field of infraredimage processing. Infrared small target is very small. And infrared small target image usually has lowsignal noise ratio (SNR), and less texture feature information. What’s worse, the background containsa significant amount of noise. Therefore, how to detect and track infrared small target robustly undercomplex background has become a more challenging research topic.Infrared small target detection and tracking system is usually divided into infrared imagebackground suppression, the infrared target detection and infrared target tracking process. In thispaper, three topics have discussed in the above. And we make a few improvements in the followingaspects.(1)We introduced the application of morphology filtering in the infrared background suppression,and proposed an new background suppression based on the soft morphology, which use weightedorder statistic to replace the maximum or minimum operation in standard gray morphology. And wecall it STop-hat algorithms. Experiment results show that the performance of this method is better thanclassical Top-hat and MTop-hat. In order to improve the adaptability of STop-hat algorithm, we putforward an adaptive structure element choosing method. Using the image mutual information curve ofTop-hat and open operate, we can get the optimal structure element size. Simulation results show that,the adaptive selection of structure element size using image mutual information is the same as themanual selection.(2)We put forward an infrared small target detection method based on the structure element size.This method can choose segmentation threshold adaptively by using structure element. Theexperimental results show that our method can well detecte small target in infrared image which isbackground suppressed by our STop-hat algorithm.(3)In the algorithm framework of the Mean-shift target tracking, we proposed a new descriptionmethod which we called triangular binary pattern, and then a new small target tracking method basedon triangular binary pattern and mean-shift has been put forward. Experiment results show that thismethod gets a better performance in target tracking than classic mean-shift methods.