The Study Of Color Image Processing Based On Mathematical Morphology

Mathematical morphology as a new image processing and analysis tools has been paid more and more attention, the application of morphology has covered many fields, including characters recognition, image compression and coding, medical image processing, vision measuring and robot vision, et al. Some image analysis systems have adopted the operations of mathematical morphology as the basic operations. The morphology method has been developed to an important research field in image processing.The basic operations of mathematical morphology are based on the set-theory and the topology. They have the integrated mathematics foundations. We can combine and deduce some applied algorithms with the fundamental operations. These algorithms can be used in the respects of image processing such as feature extraction, noise suppression, edge detection, image segmentation and so on. In this paper, we have introduced the theory of binary morphology and gray morphology and have researched the application of morphology in color image processing. The major works involved are as follows:1) Morphological filter: Based on the open and close operations, a new type color morphological filter is constructed using multiple structural elements and adaptive weighted averaging technique. The filter can not only suppress the noise efficiently, but also preserve the geometrical features of the image.2) Edge detection: Present a novel algorithm for edge detection of color image on the basis of improved edge detection operator and multi-scale structural elements. The ideal image edge under the environment of existing noise is obtained by integrating the edge characteristics of each scale element. The experimental results show that the method is effective.3) Image segmentation: Based on the watershed algorithm and region merging, a novel algorithm is proposed for image segmentation of color image. First, a gradient image is separated into a large number of small regions by watershed algorithm. Then the regions are merged on the basis of the region similarity. The algorithm reduces the excessive segmentation effectively; the image segmentation result includes more detail information; it's also more consistent with human vision properties.