Research Of The Segmentation Technology Based On The Human Slice Image

The image segmentation is classical technology of image processing, and has been always an interesting research area. The medical image segmentation is a basis of the continuous processing, such as the visualization of the healthy and diseased tissue, surgery simulation and image guided planning. The accuracy of the segmentation is vital for the doctor to make a judgment on the diseases and draw out a corresponding curing plan. Therefore, segmentation is meaningful in the medical application, especially the human slice image segmentation which is relevant to the human health, is of vital importance to visualization and the hot topic of the visualization in scientific computing. Though so many methods on edge detection, region segmentation and object extraction are put forward, there is no one that is perfectly applicable at large, and it's wise to take an effective algorithm to a certain application, so is the complex medical image. Segmentation has been the bottleneck of the development of the image processing.This dissertation studies the high resolution color human slice images of the "Visual Human Project" woman No.1, which supplied by the Medical University of Southern China. Deeply discuss and analyze on the slice image preprocessing and segmentation, also their realization. The main research and conclusion follow as:(1). It's necessary to get rid of the redundant information by special preprocessing as the huge volume and the complexity of the slice images data, then step into the further extraction. The background of each color slice images is simplified via color space transformation, and the right hand image is well segmented. The primary grey image segmentation algorithms are detailedly discussed and analyzed, which advantages and disadvantages are compared, and this thesis gives an experiment on the segmentation using some of the methods.(2). Segmentation of color image is mostly depended on the appropriate color space transformation and effective algorithms. Most of the color image segmentationalgorithms are based on the gray image segmentation development. This dissertation gives a summary and conclusion of the color space transformation and their advantages and disadvantages, and put forward a reference in the color space selection. What more, based on the color space transformation of HSI, this dissertation puts forward a new method by using an improved region grow algorithm, applying the morphology and classical edge detection operators.(3).Sometimes it doesn't perfectly work the classical segmentation algorithm, which is true for the complicate color medical image. This dissertation studies the segmentation algorithm based on the active contour model (snake model), states the drawbacks and improvement of the classical model, and makes an improvement on the greedy algorithm and comes out with a rule for the snake to terminate to the edge, which indicates that it's a preferable algorithm in image segmentation.