Research On Moving Targets Tracking Algorithm Of Passive Millimeter Wave Imaging

The Passive Millimeter wave imaging system can detect the millimeter radiation energy of the object, and the image is obtained according to the energy difference. The millimeter wave frequency spectrum distributes between infrared and microwave, it can penetrate fog, smoke and clothing to transmit. With respect to optical and infrared images, the Millimeter wave image has great advantages in object detection and tracking, which will not be affected by the weather and can detect the concealed object carried under the clothes. Besides, the Passive millimeter wave imaging system is a passive system, which needs not to emit the electromagnetic wave. So it has low radiation and good concealment. In a word, the passive millimeter wave imaging technology has great value in both military and civilian fields.Recently, Target detection and tracking technology becomes the focus of research both at home and abroad, which is widely used in the optical and infrared imaging systems but not in Millimeter wave imaging system. Because of the technology of the imaging system, the images usually have a low resolution, weak degree of differentiation between the target and the background, so the characteristics extraction of the target is not easy, the traditional target tracking method cannot be directly used for millimeter wave image target tracking. The imaging theory and the characteristics of millimeter wave images are mainly studied in this paper. According to the characteristics of the millimeter wave images, we study and improve the classical tracking algorithm. The main research contents and conclusions are shown as follows:1. According to the characteristics of the millimeter wave image, we study the weighted grey feature histogram and gradient direction histogram, then the gray level and gradient features fusion is used to model the target. The simulation result shows the fusion feature has a more comprehensive description of the target.2. The application of the mean shift algorithm in target tracking is studied, with the target modeled by the fusion feature, we make simulation to optical image and millimeter wave image respectively. The results show that the algorithm has better performance in real-time and accuracy.3. Study the theory of particle filter, which is applied to the millimeter wave image target tracking. Considering the real-time performance, we use a small number of particles but it may lead to a poor robustness of tracking. The improved immune particle filter tracking algorithm is proposed in this paper. The simulation results show that the algorithm enhance the anti-interference in target tracking without a significant increase in computational complexity.4. Study the multi-target tracking theory, especially the background reducing technical of code book model and data correlation algorithm. Considering the real-time and stability of the tracking algorithm, the improved detection based nearest neighbor particle filter tracking algorithm is proposed in this paper. Experiments for optical image and millimeter wave image in multi-target tracking respectively show that multiple targets can be tracked under the not too complicated scene. And the algorithm can handle the object occlusion situation to a certain extent.