| With the development of science and technology, digital image processing technology has more and more increasingly applied especially in military field, and played a vital role. According to military TV target tracing features, this thesis proposes a series of detailed image processing, recognizing and tracing theory after a target being locked.A TV target tracing system has following features: real-time, image indistinctness on the ground of dynamic tracing and external environmental distortion, high noise background intensity and so on. According to these features, this thesis not only presents how to tone up, segment, match after a target being locked and extract image by using Sobel brim extraction method but also raise a new thresholdselection method--fuzzy threshold method when extractingbrim.In a TV target tracing system, image matching recognition also play an important part. This thesis applies correlation and SSDA acceleration algorithm to image matching. To ensure good real-time, I first locate barycenter, then matching in comparatively small region, which increasingly improve processing efficiency.At last, I perform the test based on technology theory proposed in this thesis. The test proves that technology solution used by this thesis is right and feasible. |