| 摘要 |
The invention relates to a method of decoding coded signals corresponding to successive segmented pictures, or partitions, divided into homogeneous regions to which specific labels are respectively associated. The partitions have been previously coded by means of a coding method comprising, for each partition to be coded, the steps of translating the picture of labels into a chain in which the elements are defined by means of their movements through successive basic cells, tracking inside each cell each contour from its initial point to its end by storing symbols corresponding both to the elements of said cell and to priorities between them, and repeating these two steps up to the end of each contour of the partition and coding the information corresponding to the chains of movements. The corresponding decoding method then comprises (A) a forward decoding step, provided for decoding within said chain symbols all contour points that are sure in the decoded partition and detecting local inconsistencies, and (B) a backward decoding step, provided for tracking in a backwards manner the decoded contours and solving said inconsistencies on the basis of marks introduced during the detection of these inconsistencies. |
