Research And Implementation Of Image Segmentation Algorithm

In general, Image technique is the summary of all kinds of the skills in relation to image. There are many types in image technique, and the range of them is very wide, but they can beincluded in a holistic frame--image project. Image project is a new science which works on thewhole image field, its content is very plenty, three characteristic layers is partitioned: image process, image analysis, image understanding, which is based on different abstract degrees, research methods and so on.Image segmentation is a crucial step transiting from image process to image analysis, and holds important position in image project. On the one hand, image segmentation is the foundation of target expression, and has vita! effect on feature survey. On the other hand, original image can be translated into more abstract and more compact format by image segmentation and target expression, feature extraction, parameter survey, and so on which are based on segmentation, this makes more high image analysis and image understanding possible.Image segmentation is an important image technique, this paper has studied on image segmentation technique. The primary tasks of the paper are as follows:1 , The current and mainstream gray-level image segmentation algorithms have analyzed, classified, induced and summarized, then we gave the advantages and disadvantages of all kinds of methods, this provides definite bases for the people who choose different segmentation algorithms in the diverse application occasions and In the dissimilar data conditions.2, In these years, color image segmentation attracts more and more attentions. In the color image segmentation field, firstly an appropriate color space should be chosen, then we should use a good segmentation method corresponding to this color space. Therefore, in this paper, the color spaces, which have recently appeared in color image segmentation field, have induced and summarized, then we gave the advantages and disadvantages of all kinds of color spaces. When the people segment color images, this can provide some bases for choosing a suitable color space.3, In this paper, an efficient method in gray image two-level thresholding is proposed, which is based on the maximum entropy principle and Bayesian formula. The Gaussian distribution of theimage histogram isn't assumed in this method. Moreover, the segmented image preserves the most information of original image. Our experiments demonstrate that our method is effective and achieves a significant improvement in speed in comparison with the fuzzy entropy based on the exhaustive search method. The two-level thresholding method can be extended to multilevel thresholding. To most existent multilevel thresholding, as the number of levels required increases, the complexity and computation time will also significantly increase. However, the gray image multilevel thresholding in this paper has great superiority in speed. Detailed content is given in the fifth chapter.