Automatic Target Recognition And Tracking System Based On Template Image

With the rapid development of modem information technology, Automatic TargetRecognition (ATR) system plays an important role in the national economy and defence. Akind of template-eigenvalue based recognition and parallel tracking algorithm is designedin this paper. The algorithm focuses on the detection, recognition and tracking of movingobjects in sequences of images. Then a high-speed recognition and tracking system isestablished. Study on the system is to improve processing speed and realize the recognitionand tracking real-timely. On the basis of tamplate-matching method, some imagesegmentation and boundary detection algorithms on template-matching method are studied.And then a new Hausdorff Distance method with real-timely effect is present to recognizethe object. Simulation results are given to show the object features could be extracted fromimage sequences using this algorithm, and the object can be recognized quickly. A paralleltracking algorithm is proposed based on improved centroid tracking and template-matchingmethod. Tracking speed is improved greatly because the matching process is taken placeby the confidence factor estimation. High-speed tracking problem under disturbedsituations could be solved effectively and easily.The recognition and tracking algorithms in hardware are discussed in this paper. Thehardware platform of development is based on CCD camera. Simulation results are givento show the object could be recognized and tracked quickly. Then ADSP-BF533 hardwareplatform is introduced in this paper, which made a stability foundation for the objectrecognition and tracking system highly-speed realized on DSP.