Coding Technology Research Based On The Image Of Region Of Interest

In image processing, people can give different regard to the different region of the image. Some specific area of the still image may become the key observation, but the background which is outside this area cannot have the too much information needed. This is so-called "Region of Interest (ROI)" compression processing in the image process. When compression,these specific regions can obtain the better compression effect than that of the background, and which can be first processed in the transmission process. Based on summarizing the result of present research, the emphases of this article is the research of the extraction of Regions of Interest, and present a scheme of Region of Interest compression processing of image process.First, the characteristic of self-adaptation grads rise to search peak value is analyzed, the method of Mean Shift is applied to segment images. The method of stochastic selecting the research springboard must be used in Mean Shift, otherwise the result of segmentation could be erratic, aim at advance this disadvantage , an advanced feature space analyze is used in this thesis. at the same time, using the characteristic of self-conform of non-local means in scaling the connection of image element, it is applied in Mean Shift to avoid the disadvantage of using distance to scaling the connection of image element only. Although it increases the complexity because of execute non-local process. The experiment indicates that the result of segmentation is perfect.Second, using the gray value of edge intensity to amend the threshold, a method of self-adaptation image segmentation is advanced. This method has the advantage of little account quantity, avoided noise disturb threshold. The experimental results indicate that by using correcting to threshold by gray value can better segment image, distill the ROI.Finally, introduce Regions of Interest coding scheme based on wavelet transform,Maxshift,SPIHT, and realize it using Matlab,Method of this thesis can better advance quality of reconstructed images compared with ROI compression coding method using SPIHT.