Method and device of the spatial filtering of images decoded by transformation by unit.

摘要

Digital image coding and decoding by pixel block transform with a view to reducing the information throughput leads to discontinuities of brightness level for fringe pixels (D, E) at the boundaries of blocks (BL1, BL2). These discontinuities are detected for each fringe pixel (D) through a change of sign of two gradients of level which are calculated respectively with the levels of two pixels (C, E) adjacent to the fringe pixel (D). The brightness level (NL) of the fringe pixel (D) is then corrected through a discontinuity correction factor (FCD) evaluated from a deviation of discontinuity (ED) equal to the absolute value of the difference of the brightness level of the fringe pixel (D) and the arithmetic mean of the brightness levels of the adjacent pixels, dependent upon a quantisation step associated with the block (BL1). IMAGE