Macroblock Skip-mode Prediction for Complexity Control of Video Encoders

In this paper we propose a macroblock skip-mode prediction algorithm to reduce the computational effort of video encoders. The algorithm classifies each macroblock as “skipped” or “not skipped” by estimating the energy of low-frequency quantized coefficients prior to coding, making it possible to significantly reduce computation by not coding these skipped macroblocks Results show that the algorithm can achieve substantial computational savings with only a small degradation in rate-distortion performance.