| 摘要 |
PURPOSE: A method for compressing fractal images is provided to improve image compression rate and process the edge and texture of an image well. CONSTITUTION: A processing block is transformed by A transform using bits of T(b,R)/4 to search the original image for a similar block in size of 2bX2b satisfying error allowance. When the similar block has been found, A transform is stored corresponding to the processing block without increasing the number of bits. When the similar block has not been found, the processing block is transformed by B transform using bits of T(b,R) to search the original image for the similar block. When the similar block has been found, the B transform is stored corresponding to the processing block. If the similar block has been found, the processing block is divided into four to generate four sub-blocks each of which has the size of b/2Xb/2. The sub-blocks are A-transformed or B-transformed to search the original image for a similar block until the size of the divided block becomes 2X2. |
