High-performance adaptive group-of-picture rate control for H.264/AVC

Abstract

In this paper, an adaptive group-of-picture (AGOP) control is proposed to improve the coding performance for the H.264/AVC system. The efficient algorithm is developed to find the quantized parameter with temporal correlation between inter-frames. According to the estimated parameters, the GOP coding type can be adaptively selected as AGOP or BGOP (basic GOP). The rate control with some decision functions is proposed to modify the quantization parameter (QP). The coding rate can accurately meet the target rate with adjusting QP value dynamically, while keeping low computational overhead. Simulation results show that the proposed algorithm can improve PSNR by about 1 dB in average compared with the original H.264/AVC reference software (JM12.4), from various sequences testing.

Temporal information based GoP adaptation for linear video delivery schemes

“…In addition, the PSNR scores are computed to evaluate the PSNR fluctuations inside each subsequence. Indeed, numerous studies [20,37,38] reported…”

“…By optimal, we mean the GoP-size that optimizes the trade-off between received quality and complexity cost. Despite, some adaptive GoP-size solutions for conventional standards (e.g., H.264/AVC, H.264/SVC) or Wyner-Ziv codec already exist [19][20][21] they cannot be directly applied to SoftCast. Indeed, different from conventional approaches that rely on motion estimation/compensation and entropy coding, SoftCast uses a temporal DCT to exploit inter-frame correlation and a pseudo-analog modulation for transmission.…”