This paper presents a new efficient scheme for lossless compression of binary images, by coding their morphological skeleton representation. Computer simulations indicate that, typically, the proposed coding scheme significantly improves the coding rates obtained by the best previously known schemes for skeleton coding, and is substantially more efficient than coding the original binary image by chain code, quadtree and run-length/Huffman methods. Comparison to existing coding standards for scanned bilevel documents places the proposed algorithm between the Group 3 and Group 4 algorithms, in terms of compression efficiency. The proposed algorithm is fast, when properly implemented. The proposed scheme is based on new theoretical results obtained by the authors concerning properties of the morphological skeleton representation.
展开▼