Research On Infrared Moving Target Detection Technology Based On DSP

With the development of infrared technology, with its ability of good anti-interference, working all day, good concealment, infrared detectors have wide range of applications in the military and civil fields. Infrared moving target detection is an important branch of infrared technology, which provides a necessary preparation for the follow-up steps, such as target tracking, target recognition, image analysis and so on. This is a project of great significance.In this paper, we introduced the characteristics, composition and processing of infrared imag, discussed and studied three kinds of detection algorithms in detail, which contains optical flow method, frame difference and background difference. Through the analysis we understand that the frame difference has a better dynamic adaptability and has the advantages of simple and practical. But the detected target may have some holes inside of it. While the general background difference is different, it can extract the moving target more completely, but the ability to resist the interference is poor. For the background difference, we can improve the final results through the appropriate background modeling and updating. In this paper, the target detection processing algorithm mainly consists of four parts, including:(1) the three frame difference method,(2) Surendra background modeling, 3) Gaussian background modeling, when the color of object is close to the background, Gaussian background modeling may not extract the complete contour of object. So in the process of visible light video, we can combine the three frame difference method with the Gaussian background modeling. In the processing of infrared video, we firstly make the image histogram equalization and then carry out the Gauss model processing.(4) In the paper, we proposed a method that only processing the moving regions using frame difference method or background subtraction method. Through the frame differenc of moving regions, we verified that the method can reduce the computation quantity while guaranteeing the effect. All of the four kinds of algorithms show good results in the processing of infrared video and visible video.In order to meet the requirements of the project, we need to transplant the algorithm to the embedded platform DSP to realize the offline processing. Here are the two development board, ADI Bf548 and TI DM6437. The whole process is divided into video capture, video processing, the results display module. The whole process and the design are consistent with the processing of the software platform. The video processing includes image gray processing, background subtraction, threshold segmentation, tracking the targe using a square. In the end, experiments showed that we achieved the detection and tracking of the visible and infrared moving target detection successfully...