Research On The Improvement Of Local Code Book Based On Fractal And Image Weighting Algorithm

With the rapid development of the Internet and multimedia technologies,the amount of image data that computers need to process is rapidly increasing.In order to efficiently handle these large amounts of data,various image databases and related processing algorithms have been generated.Image retrieval is one of the core technologies in the field of image processing.It is to be a hot topic that how to quickly locate a certain image in the data set for people to solve.The main research work of this thesis as follows:1.A new image retrieval algorithm based on quadtree partitioning is presented.The local codebook is used,and the selection range of the codebook is limited to the sub-blocks.A variety of local codebook generation strategies are implemented,and the local codebook with the highest PSNR value after decoding is selected,and the image decoding quality is improved under the premise of guaranteeing the encoding speed.In the image distance formula,according to the different position relationships between sub-blocks,weight information is added in its corresponding distance formula,thereby improving the performance of the search algorithm.2.A fractal image retrieval algorithm based on triangle segmentation is presented.Two different triangles are used to segment the image,which improves the adaptability of the algorithm to the image.The four kinds of D_i blocks around the R_i block are selected as the codebook set,which reduces the number of codebooks and improves the coding speed of the coding algorithm on the premise of ensuring the image decoding quality.3.An image distance comparison formula is proposed,which used the non-overlapping area value between different codebook blocks as the distance and weight between sub-blocks,further improve the search algorithm performance.Experiments have verified the effectiveness of the given method.