Realization And Design Of System Of Re-entry Optic Multi-target Detection And Recognition

The implementation of the automatic target recognition by TV tracking systemdepends on some vital video processing technology, such as the restrain of backgroundand noise in video image, the segmentation and the characteristic extraction of image,and the recognition and tracking of the moving target. For the close target, the analysisof texture, moving, shape, and mode is needed. For the distant target, the methods ofarea detecting, image segmentation, etc. are different from the common target.TV tracking system is the principal part of the automatic target recognition system.In this paper, as basic knowledge, video signal processing, digital image processing,threshold calculation were analyzed to present the signal detection principle. There areseveral key technologies to realize the function of automatic recognition and trackingthe selected target of re-entry complicated multi-target. These technologies include fainttarget processing, multi-target processing, real-time image processing and extraction oftarget characteristic.According to the special requirement of automatic target recognition, this paperanalyzed the priority of track before detection. Images sequence was pre-processed toremove the noise, enhance the contrast between background and target. The imagesegmentation based on changed area detection was combined with image segmentationby bidirectional and perpendicular extraction of moving multi-target based on leastdistance. Auto-adaptive digital gate system was devised, and the method of the judgingthe central position of target by the gate was discussed. At last, trajectory associatingwas applied to identify the tracked targets. These technologies set the base for theautomatic target recognition.For the complicated physical phenomenon of re-entry targets, the optic trackingsystem was unable to recognize targets correctly. In this paper the characteristic of flareand moving of all kinds of re-entry targets were analyzed carefully. There are someuseful conclusions. The time and intensity of different flare targets are different. Thedimensions of targets are different. The positions change can be tracked. Speeds ofdeparture between targets are different and speed change of some targets is stable. All of these conclusions are very helpful for the recognition of warhead.Based on the image segmentation and characteristic extraction, this paperresearched and found the method of automatic target recognition by the segmentedimage. Firstly, the distinct characteristics are applied directly, avoiding noise affection.Then, with light-flow equation, the extracted characteristics are tracked, and the movingof target is predicted by Kalman filter equation. At last, according to the matching of therelative positions, dimensions, moving directions and relative speeds, the opticcharacteristics and moving characteristics are fused to implement the automatic targetrecognition for the complicated optic multi-target.This paper discussed the layered design method, according to the basic function ofautomatic target recognition. For different demand, two kind of hardware system aredevised, one is improved system based on existing equipment, the other is fullyreconstructed, and the processing modes of software are analyzed too. The test in thefield proved that this system is robust and stable, and function correctly in therecognition and tracking target. It’s very important for the test of missile.In conclusion, it is urgent demand for the missile test to implement multi-targetautomatic identify and tracking. It is also the future progress of intelligentization ofoptic equipment. This paper is supported by corresponding project, and has acomprehensive appliance. It has a reference for the other fields of video processing.