Two-Pass Rate Control for Improved Quality of Experience in UHDTV Delivery

Abstract

Rate control plays an important role in any video coding application and it was extensively studied in the context of previous video coding standards. However, the current state-of the-art high efficiency video coding (HEVC) standard introduces many flexible tools making previous rate-distortion models used in rate control insufficiently accurate. Recently, a few rate control methods have been developed for HEVC that introduce many useful features, such as a robust correspondence between the rate and Lagrange multiplier λ. Nonetheless, previous rate control algorithms for HEVC do not address typical content in television applications that consists of frequent scene changes. Furthermore, the new ultra high definition television (UHDTV) format, which is expected to become widespread in the future, demands for even higher compression efficiency. To overcome these issues, a two-pass rate control method is proposed in this paper, targeting the encoding of UHDTV content. In the first pass, a fast encoder with limited set of coding tools is used during pre-encoding step to obtain the data used for rate allocation and model parameter initialization, which will then be used during the second pass. To avoid multiple encoding steps when deriving this information, a variable quantization parameter framework is proposed. Experimental results show that the proposed rate control method outperforms the well-known HEVC rate control method. When compared with variable bit-rate encoding mode, the proposed two-pass rate control method achieves on average 2.9% BD-rate losses. That is significantly better than the state-of-the-art HEVC rate control method, which achieves an average 8.8% BD-rate loss. The proposed method also provides a more consistent quality fluctuation with time, measured with standard deviation of frame PSNR values, required for high Quality of Experience.