Morphological reconstruction of semantic layers in map images

Map images are composed of semantic layers depicted in arbitrary color. Often images require separation into layers for storage and processing. Separation results in severe artifacts because of layer overlapping. In the current work, we design the technique to restore semantic layers after separation. The designed restoration technique alleviates compression deficiency of reconstructed layers versus corrupted ones with lossless compression algorithms (ITU Group 4, PNG, JBIG and IBIG2 with optimized context templates), and provides better visual quality of images in operations with selective layer removal/extraction.