An improvement to fine granularity scalability based on H.26L

FGS scheme has received much attention for its advantage of adaptation to various bandwidths. Incorporation of FGS into recently emerged H.26L can further improve its adaptation to the network communication. In this paper, we propose an improvement, which contains two parts, to the original FGS encoding scheme. First, we use run coding of the skipped macroblocks/blocks in the most significant bit planes. This is due to such fact that generally there are only few macroblocks/blocks encoded in these bit planes. The experimental results show that it can greatly reduce the coded size of some bit planes, up to more than 90% for first bit planes. Second, for the least significant bit planes, another improved coding method, adaptive pixel-by-pixel coding, was developed. It can save up to 14000 bits in the first LSBP and 3% of whole FGS size. Meanwhile, the increase in complexity and memory requirement are both negligible.