Research Mathematical Morphology And Watershed Algorithm For Sensing Remote Image Object Recognition

Image recognition technology is extremely active research subject in the area of the computer image processing and pattern recognition. The watershed algorithm and mathematical morphology algorithm has the widespread application in image object recognition. This paper unifies these two methods, and uses in remote sensing image object recognition. The major work summarized as follows:(1) Discusses the principles of watershed algorithm and its application in image processing. Watershed algorithm is a segmentation based on the region character, which may carries on the operation to the original image and the gradient image, but the watershed algorithm is easy to have over-segmentation. It's produced a method to pre-process,then uses region-growth watershed algorithm to extract road information.(2) Discusses mathematical morphology. It's proposed a image object recognition algorithm based on mathematical morphology and watershed algorithm. The method uses morphological scale space to smooth the original image, and the threshold of gradient to optimize the image, then uses watershed algorithm .The experiment of remote sensing image shows this method is efficient.(3) Discusses the rough set theory in detail. Focus on the algorithm combining of the mathematical morphology and rough set. Firtly uses the rough set filtering to process the image, then uses morphology to extract the edge of the image, finally uses the geometric characteristics of image to remove the area which we don't need. The experiment of remote sensing image proves its feasibility.