Efficient Lagrange multiplier selection for scalable video coding using temporal inter-layer dependency

Designed for scalable video coding, an efficient algorithm for Lagrange multiplier selection, named as Lagrange multiplier selection with temporal scalability, is proposed based on inter-layer dependency in the temporal domain. The inter-layer dependency between the temporal base layer and higher temporal layers is explored and used to select the Lagrange multiplier, which leads to efficient allocation of bit budgets to different temporal layers according to their impact and importance in optimization of the overall video quality. Experimental results show that compared with using the original Lagrange multiplier in the reference software model of H.264/SVC (i.e., JSVM), the proposed Lagrange multiplier achieves a performance gain of up to 0.41 dB in Delta PSNR and 10.61 % in bit-rate reduction at various bit-rates. At lower bit-rates, the performance gain can be further improved. Consistent performance gain is also observed when compared with two other state-of-the-art Lagrange multiplier selection techniques for scalable video coding.