Abstract:
In this paper, based on MPEG-2 DBN networks to H.264 fast transcoding algorithm, using H.264 macro block (MB) mode selection and mpeg-2 motion compensation, the correlation among the residuals, the H.264 macroblock mode selection into data classification; In MPEG-2 decoder, save the related information, macro blocks including MB coding mode, coding block types (CBPC), MB residual mean and variance of decoding using standard H.264 encoder to YUV image coding, save the H.264 macro block coding mode, to get classifier with deep learning algorithms, used for H.264 coding mode classification, solved the H.264 macroblock mode decision problem, reduces the traditional method for macroblock mode decision of the high computational complexity, thus greatly improving the efficiency of transcoding.
Tag:
点此返回栏目查看更多>>>参考论文