Source area extraction is one of the most important tasks in IC (Integrated Circuits) image processing. It is a problem of image segmentation on specialized images whose special characteristic must be taken into account during segmentation to ensure an accurate and fast result. We present tow new algorithm based on analyzed the character similarity as well as difference between that of source area of IC images. We present a Image Segmentation algorithm based on Watershed and prior knowledge. Frist of all we transform the original image to watershed then we calculate the energy of the label image result from the Watershed transform by designing a prior density that penalizes the area of homogeneous parts in images . The segmentation problem is the maximizing a posteriori estimation（MAP） of the set of areas such we can find the optimal areas of object, and the other areas of the image are looked as background areas . The experiments indicate our Algorithm is effective for IC image segmentation. Then we extend this algorithm into RGB space. In some of the color images didn’t adapt to process after transforming to gray image for it will lost some important information. We use the color image as original image transform to watershed image, than we use a news 3-D prior knowledge find the optimal object area. This algorithm not only solved the oversegmentation problems of watershed transform ,but also used color information and prior knowledge. The experiments indicate our Algorithm is effective for image segmentation. In the last Section ， we present a new texture segmentation base on image decomposition . we use a new textural feature and smooth feature as feature channel based on image decomposition technology,then we use them in a unsupervised texture segmentation method.
修改评论