| 摘要 |
PURPOSE: To provide an image processing method which reduces the redundancy by making use of the correlation between adjacent pixels and can improve the image compressibility. CONSTITUTION: An original image A is subsampled in 9 pieces. A mean value image B' consisting of the mean pixel value of those divided s bsampling images B1 to B9 is obtained together with the differential images B'1 to B'4 and B'6 to B'9 which has the differences from the image B'. Then the image conversion is carried out for reduction of the redundancy. The image B' is subsampled again in 9 pieces, and the image conversion is carried out in steps for reduction of the redundancy. Thus the final mean value image B" and the phased differential images B'1 to B'4, B'6 to B'9, B"1 to B"4 and B"6 to B"9 are quantized and encoded in variable length. |
