Complexity-Aware Frame-Level Bit Allocation and Rate Control for H.264 Scalable Video Coding.

This study aimed to investigate efficient bit-allocation and rate-control methods for H.264 scalable video encoders. We first presented a new bit-allocation algorithm at the frame level for H.264 temporal scalability. Additional to the target bit-rate and buffer occupancy, the temporal hierarchical level and visual complexity were considered for determining the bit budget of a frame. This bit-allocation scheme was followed by a quantization parameter (QP) estimation procedure to achieve the desired bit-rate. We improved the conventional QP assignment by identifying and correcting inappropriate QP values due to incorrect visual complexity prediction. Furthermore, the proposed bit-allocation and rate-control algorithm for temporal scalability was seamlessly integrated with spatial/quality scalabilities to offer flexible bit-rate/quality combinations. Rate-control performance of an investigated method was extensively evaluated in the accuracy of produced bit rates, rate-distortion behavior, and stability of the decoder's buffer occupancy. Experimental results showed that the proposed algorithm achieved accurate bit-rates, prevented buffer overflow and underflow, and simultaneously produced more favorable and stable visual quality as compared to the state-of-the-art approaches in the literature. [ABSTRACT FROM AUTHOR]