Lossless Multi-Grid Chain Code with Optimized Cell Usage for Contour Description of Visual Objects

In the context of object-based image compression, shape encoders that describe the contour of the objects, called chain codes, tend to be more efficient than that the one based on bitmap, which is the MPEG-4 Part 2 framework. Among the chain codes, those based on grid cells are expected to outperform the ones that employ a symbol to code each particular boundary pixel. This paper presents a lossless chain code based approach for binary shapes encoding, called OCULMGCC (Optimized Cell Usage - Lossless Multi Grid Chain Code). The OCU-LMGCC solves shortcomings in the original MGCC, optimizes the required extra parallel information to the decoder and uses a new set of coding symbols, better suited to the novel cell-switching schemes proposed. Results showed a compression gain noticeable larger than the known algorithms reported in the literature.