Method for processing images by regions, using discrete transformation on finite segments without extension

摘要

The method involves an image (I) being fragmented into regions (R0,R1,R2). A region (R1) of the image is split into segments. Each segment is formed from a sequence of points (n0,n1-n14) corresponding to the contours (Z) of the region. Digital samples of the points of the segments for the image region are collected. Components of approximation and the detail components corresponding to a central part of the segment are obtained. A discrete transformation is performed on the image samples corresponding to the segment. The method involves calculation of the approximation components of the segment extremity parts, by extrapolating the approximation component values from the central part of the segment. The detail components are deduced from the extremity parts of the segments, from the approximation components calculated by extrapolation and from the samples corresponding to the extremity parts of the segment, to permit a perfect reconstruction of the samples of the image segment.