Automatic target detection is very difficult in complicate background of sea and sky because of the clutter caused by waves and clouds nearby the sea-level line. In this paper, in view of the low-flying target near the sea is always above the sea-level line, we can first locate the sea-level line, and neglect the image data beneath the sea-level line. Thus the waves under the sea-level line can be suppressed, and the executive time of target segmentation is also much reduced. A new method is proposed, which first uses median value filtering to suppress background and enhance targets so as to increase SNR, then uses a method of the multi-point multi-layer vertical Sobel operator combined with linear least squares fitting and a method of sea-sky background segmentation based on row mean value to segment the sea-sky background, lastly uses the binary-centroid tracking algorithm to detect and track the target.In the experiment, high frame rate and high-resolution digital CCD camera and high performance DSPs are applied. We take full advantage of DSPs's characteristic of high performance in order to resolve problem of large amounts of... |