Based Matching Algorithm For Multi-chip Dsp

In modern war, air-to-air missile shows the important ability, and is the important guarantee of obtaining the victorious war. In air war, infrared control and guide air-to-air missiles are used primarily. The related tracking algorithm is the key technique of infrared control and guide air-to-air missiles. The precision of the related matching algorithm is very high, but the operand is big. As a result, it can't get the extensive application in real-time image matching field.Firstly, a two-stage image matching algorithm based on SUSAN and image separating and template separating method was developed, in which coarse searching and accurate location are combined. Image was separated. Tracking template was confirmed after by taking advance of SUSAN. Then template was separated. Matching point was confirmed through child template common matching peak value point and through coarse searching and accurate location. matching point was analyzed by a new creditability evaluation method. Reserving template and giving invalid template.Secondly, a serial of picture in which contrast of target is decreased, and the variety of average gray is big is simulated. As a result, this algorithm can realize precise orientation, and shows excellent tracking performance, and effectively increases tracking accuracy. When a part of the picture is vary, a two-stage image matching algorithm based on SUSAN and image separating and template separating method can estimate modified position (final matching position) based on matching position of 4 child templates, and orientate the accurate target, and effectively decrease matching performance which is caused by geometric distortion of big template. At last, realizing the method of the algorithm way is described by multi -DSP in detail. TMS320C6416 which is the quickest DSP signal processor is selected. It can work at 720MHz. Its clock period is shortened by comparing other DSP. The production time of related matching is decreased at large. It can meet the demand of the real-time image matching system.