CPC H04N 19/172 (2014.11) [H04N 19/176 (2014.11)] | 16 Claims |
1. An image compression method, comprising steps of:
acquiring a human-eye fixation point on an original image, and determining a fixation region and a non-fixation region of the original image according to the human-eye fixation point; and
compressing the non-fixation region by dividing the non-fixation region into a plurality of image blocks which do not overlap with each other, and rearranging the plurality of image blocks to generate a rearranged region; and
compressing the rearranged region by down-sampling a chrominance channel of the rearranged region in a first direction according to a preset first sampling multiple; and
down-sampling a luminance channel and the chrominance channel of the rearranged region in a second direction according to a preset second sampling multiple;
wherein each image block is rectangular, two of its four sides are parallel to a horizontal direction, the other two sides are parallel to a vertical direction, the first direction is one of the horizontal direction and the vertical direction, and the second direction is the other of the horizontal direction and the vertical direction; the first sampling multiple is greater than the second sampling multiple; and
generating a compressed image according to the fixation region and the compressed non-fixation region.
|