Adaptive Quantization for Interlaced Video Coding

This paper presents an adaptive quantization method by adjusting the rounding offset for scalar quantization adopted in AVS 1.0 part 2 zengqiang profile. The rounding offset is chosen adaptively based on minimum transform coefficients energy difference for quantization, and the transform coefficient energy difference is calculated by the sum of absolute distortion before quantization and after dequantization simply. The proposed adaptive quantization provides up to about 0.3 dB improvement for high bitrate application in terms of peak signal-to-noise ratio (PSNR) based on public AVS reference software, with low computational complexity and memory consuming, especially for the interlaced video coding