Modeling the bit rate of H.264/AVC video encoding as a function of quantization parameter, frame rate and GoP characteristics

We present a novel analytical bit rate model for H.264/AVC video encoding which is based on the quantization parameter, the frame rate, the group-of-pictures length, the group-of-pictures structure and content-dependent parameters. The content-dependent parameters are described as a function of two standard video activity measures (temporal and spatial activity) which can easily be determined from the uncompressed video. Our bit rate model shows a Pearson correlation of more than 0.97 and a root-mean-square error of less than 2% as determined by statistical analysis.