Research On Moving Target Cooperative-Detection Under Complex Background

Moving target detection technology in image sequence is one of the most important content in computer vision, and be widely used in robot visual, intelligent monitoring and vehicle intelligent auxiliary drive system.Traditional video detection technoldgy is restricted by the photography mechanism and the spectral feature of single-sensor, and is hardly satisfied by real work setting, so the video detection technology based on multi-sensor has been the hotspot of nowadays. With the sponsoring by the Basic Research on National Defense, this paper focuses on the study of multi-sensor cooperative-detection. The main tasks are as follows.1. The technology of single-sensor target detection, multi-sensor fusion detection and multi-sensor cooperative-detection is compared and analyzed. The base requirement of cooperative-detection technology is described, and the technology of cooperative-detection should be satisfied in real time, synchronism, interactivity and robustness.2. After finishing the architecture of the experimental system, a fast matching algorithm is adopted aiming at the fault of the matching in the infrared image and the visual image. The strategy from coarse matching to fine matching can achieve the job of image matching and the target location. Characteristic dot pair by coarse matching is used in estimating the transformation parameter, and using the finest parameter as the real transformation parameter of the images can achieve the job of fine matching. Experiments show that the algorithm is effective.3. In the pre-processing of image sequence, spatial filtering technology and space time filtering technology are recommended. In the target detection of single-sensor, aiming at the infrared image and the visual image, the different target detection methods based on adaptive threshold segmentation algorithm and multi-frame differencing are put forward in this paper. Experimental results show that the algorithm can be used for kinds of objects detection effectively, with high real-time and robustness performance.4. Aiming at the fault of the detection result automatic evaluation, a method based on evaluation factors and gray relation is put forward in this paper. After finishing the mathematical description of evaluation factors, several quantificational evaluation components can be gained, and then, through gray relation, the quantificational and synthetical evaluation can be gained. Experiments show that the algorithm is...