Image Segmentation Based On Mathematical Morphology

Image segmentation is vital image processing technique and the main problem of computer vision. It is also a key step from image processing to image analysis and holds significant place in image engineering. On one side, image segmenation is the basic of target expression and affects feature measure greatly. On the other side, original image can be translated into more abstract and more compact format by image segmentation and some factor based on segmentation, such as target expression, feature extraction, parameter measure, and so on. In image applications, the target extraction and measure can't be done without image segmentation. So, there are so many scientists who work on this problem and derive a lot of segmentation algorithms that are various degrees of success of different image.Segmentalizing image by mathematics morphology is the main goal of this paper. So, in this paper, we introduce origin of mathematics morphology from binary morphology to gray morphology and extensively study its diferent operators and quality. Then the image segmentation based on edge detection with morphology and region segmentation with morphology(watershed segmentation) are expounded in detail.On the side of edge detection with morphology. Considering the advantages of mathematical morphology exploited in image edge detection, a multi-scale morphological edge detection algorithm is proposed. The experiment results show that the edge detection algorithm has a good performance of noise reduction compared with traditional edge detectors, and the detected edge is smooth.On the side of region segmentation with morphology. A algorithm of watershed segmentation based on hierarchical maker extracting was proposed, this algorithm can restrain the phenomena of over-segmentation well and can obtain good results compared with traditional watershed segmentation. At the same time, we proposes a algorithm for artificialtarget segmentation from natural background based on fractal theory and hierarchical maker extracting watershed segmentation. The experiment results show that this algorithm can extract the target from natural background effectively, and keep complete contour of the artificial target as well.