Use of content complexity factors in video over IP quality monitoring

We propose to use two parameters, the scene complexity (C) and the level of motion (M), to capture the video content characteristics that matter most to the quality of compressed videos transmitted over IP networks, either with or without packet losses. Two simple but robust formulas are designed to calculate these two parameters from the following variables: bits of coded Intra (I-) frames (BitsI), bits of coded Inter (P-) frames (BitsP), the average quantization parameter of I-frames (QPI), and the average quantization parameter of P-frames (QPP). We demonstrate a simple method to incorporate the scene complexity and the level of motion to improve the performance of a video-content-blind video quality model for no packet loss case. Its expansion to quality models with packet loss is currently under study.