Video Rate Control Using Activity Based Rate Prediction

In this paper, an efficient rate control algorithm based on rate prediction is proposed for maintaining a smooth buffer variation and a small buffer size. The proposed method adjusts the quantization scaling factor by using the predicted bit-rate to meet the target bit budget exactly. Experimental result show that the proposed prediction-based rate control scheme can regulate the bit-rate across scene changes more effectively and achieve better PSNR performance than existing rate control mechanisms such as the MPEG-2 Test Model 5 (TM5) and the Adaptive Scene Analysis (ASA).