Reducing computational costs of H.264/AVC for video codec in mobile devices

Wei Yao, Zhengguo G. Li, Susanto Rahardja · 2013

In video delivery system, it is known that early detection of all-zero quantized discrete cosine transform (DCT) coefficients can be used to reduce the computational cost of H.264/AVC video encoding, which is critical for mobile applications. In this paper, a new method to efficiently and accurately detect such all-zero coefficients is presented. The early detection criterion is derived from a new rate - quantization model that analyzes the distribution model of DCT coefficients and takes into account the unique features of H.264/ AVC quantization process. Simulation results show that the proposed method consistently achieves a very high detection rate, while maintaining a low false detection ratio and negligible quality degradation. A comparison with other existing methods shows that by considering the unique features in the H.264/AVC quantization step, the detection of all-zero DCT coefficients is improved. The investigation on encoding process runtime also indicates a significant time saving by enabling the proposed early AZB detection especially for low bit-rate applications in mobile devices.

Read the paper · More papers on PaperTik