| 摘要 |
PURPOSE: A method and a device for an image compression using a wavelet transformation are provided to improve an image quality of wanted part of an image, a region having a middle luminance value, and a boundary region by quantizing an image sample according to a character of image data through an application of the fuzzy logic reflecting the human sense of sight system. CONSTITUTION: A device for an image compression using a wavelet transformation includes a BDWT mean(510), a block rearrangement mean(520), a quantization processing mean(530), and an entropy coder. The BDWT(Block wise Discrete Wavelet Transform coder) mean(510) rearranges four sub-band every level at a video frame. The block re-arrangement mean(520) divides inputted source image data into plural 8 by 8 image data, and rearranges the plural 8 by 8 image data. The quantization processing mean(530) includes a weight distribution value output part(532), a relevance output part(534), a quantization matrix output part(536), and a quantizer(538). The weight distribution value output part(532) outputs a distribution value of each block of rearranged image data output by the block re-arrangement mean(520). The relevance output part(534) outputs a relevance value from the distribution value and an entropy value of each block by applying a fuzzy logic. The quantization matrix output part(536) outputs each coefficient value of a quantization matrix from the relevance value. At this time, the number of bit assigned to each transformation coefficient is output by dividing each transformation coefficient into each coefficient of the quantization matrix, and the quantizer(538) attaches the number of bit to each transformation coefficient. The entropy coder(540) receives the output of the quantizer(538), and assigns more number of bit to a region having a middle luminance value and a boundary region of video image data by subdividing transformation coefficients of the region having the middle luminance value and the boundary region of video image data. |
