Color Medical Image Segmentation Techniques

The Medical image is used in medical diagnosis widely, so, the image processing technology plays an important role. And the imge segmentation is the based work of image process. This article uses edge detection and threshold value segmentation to segment gray-scale image. The research of the color image segmentation includes HSV color clustering and watershed segmentation. This article quotes the concept of weight, and uses it to improve the algorithm.Most of image segmentation technology is used to segment gray-scale image. This article chooses two methods which are edge detection and threshold value segmentation to segment gray-scale image. In edge detection, it introduces sobel operator and lapalcian operator. This article introduces and implements three methods:the variance between two classes threshold segmentation,amximum entropy threshold segmentation and iterative threshold segmentation. After these segmentation, some of the target is incorrectly classified as bacground,or reverse. To this problem, the idea of combining median filter and erosion of morphology is proposed. Experiments show that the proposion makes good results.As the mature of technology, color image segmentation is put on the agenda. Because color image can supply more information compare with gray's. It has more difficult to segment color image. This article introduces four commonly used color space, and uses color slice images of human to transform in space. This article implements color image segmentation in RGB space and HSV color clustering algorithm.Because of the complexity of medical images, article further study watershed segmentation method and HSV color space clustering algorithm to segment color images, and does experiment using color slice images of human. But watershed segmentation produces over-segmentation. To solve this problem, article quotes the concept of weight, and makes improvents with threshold value segmentation and HSV color space clustering algorithm separately. And they obtain good results.