Images And So On A Number Of Applications In Image Processing

In modern society, people have more and more requirements in media information exchanging, and hope get better information exchange quality. Image is the main carrier of media information. Image segmentation is the base and hot topic of image processing and computer vision field. Block effect is the main problem, which influences the image quality in image compression with low bit rate.This thesis is based on the Intel project of "audio, image/video compressing and their transferring performance research". It aims at offering effective and high quality module function to end user of "Intel Integrated Performance Primitives". The modules were realized through image segmentation and compression algorithms that based on analogies thinking. After all, customers are more convenient to develop applications. The main expatiates and works of the thesis are as follows:1. The Image Analogies algorithm and its realization were researched, and the application field of analogies thinking also extended in the thesis.2. Image Segmentation algorithms, were deeply analyzed, and the advantages and disadvantages of these algorithms were given.3. Based on analogies thinking, the image segmentation algorithm IABS have been designed. Training set is structured from images, which will be segmented. The segmentation process is, first to learn the mapping relationship/between the pair of images in the training set, and then use the relationship to segment the target image.4. The performance evaluation method of image compression encoding system is introduced. The reason of blocking effect existing and blocking effect measurement were analysed carefully, introducing the deblocking algorithm and its advantages and disadvantages.5. According to the analogies thinking, a way for image deblocking is implemented.That is the IABD algorithm, used for deblocking of the fractal compressed decoding image based on the analogies thinking. There are two steps in the process. First, we filter the waves to preprocess the input image, and get the indefinite version of the input image without blocking effect. Then we use a self-study analogies algorithm SIA to recover the particulars of the image. The high definition image and the degenerate image (the indefinite version of the original image) constitute the training set.6. The performance validation of new algorithm proved that it could get satisfied results.The image segmentation algorithm and deblocking algorithm have been realized by coding, and also have been integrated into Intel Integrated Performance Primitives for testing. By now, the algorithms execute more effectively with good experiment results. After passing all the tests, these two algorithms will greatly help developers who use "Intel Integrated Performance Primitives" on their relative projects development.