Performance improvement of HEVC using Adaptive Quantization

High Efficiency Video Coding (HEVC) is the most recent video compression standard that achieves higher encoding efficiency over earlier popular standards like MPEG-2 and H.264/AVC. By adopting a variety of coding efficiency enhancement and parallel processing tools, HEVC is in a position to provide up to 50% more bit-rate reduction over its precursor H.264/AVC. In an HEVC encoder, transforms are applied to the residual signal resulting from inter or intra-frame prediction. The transformed coefficients in a Transform Unit (TU) are then equally quantized based on the value of Quantization Parameter (QP) to provide the input for the entropy encoder. The quantization process does not consider the directional bias on the energy distribution of transformed coefficients and the scanning pattern that follows the quantization stage. This paper presents an Adaptive Quantization technique that adjusts the level of quantization based on intra prediction mode, type of component(luma or chroma) and the block size. The proposed method provides an average BD rate reduction of -0.35% and -0.46% along with a BD PSNR improvement of 0.065dB and 0.038dB for Intra Main and Intra Main 10 configurations respectively, without Rate Distortion Optimized Quantizer (RDOQ).