Abstract: AR-FGS (Adaptive Reference for Fine Granular Scalable) adopts leaky prediction to achieve a better tradeoff between coding efficiency and robustness. The leaky factor is crucial to the performance of leaky prediction. This paper proposes an adaptive method for determining the optimal leaky factor for each frame. First, the disadvantage of the bit-stream extraction method used in the current JSVM (Joint Scalable Video Model) is analyzed that the bit-rate of extracted sub-steam is not smooth at the frame level. To guarantee the smoothness, a modified extraction method is presented which truncates each frame at the fixed total bit-rate. The proposed leaky factor determination algorithm sets the optimal leaky factor for each frame according to the ratio of current reference frame’s base layer bit-rate to that of the first I-frame’s. The optimal leaky factor is further adjusted according to the ratio of several previous frames’ average base layer bit-rate to that of current reference frame’s. Simulation results show that the proposed algorithm can further improve the PSNR over a wide range of bit-rate, compared with the use of the fixed leaky factor. Additionally, the bit-rate of sub-stream extracted by the modified bit-stream extraction method is smooth at the frame level.

Key words: video coding, AR-FGS, leaky prediction, leaky factor, bit-stream extraction