Efficient coding and mapping algorithms for software-only real-time video coding at low bit rates

This paper presents efficient coding and mapping algorithms that lead to a significant speed improvement in low bit rate H.263/H263+ video encoding wh ile maintaining high video-reproduction quality. First, by exploiting stati stical properties of low resolution and slowly varying video sequences, we reduce significantly the computation times of the most computationally inte nsive components of video coding, particularly the discrete cosine transfor m, the inverse discrete cosine transform, quantization, and motion estimati on. We also map some of the single instruction multiple fata (SIMD)-oriente d functions onto Intel's MMX architecture. The developed algorithms are imp lemented using our public-domain H.263/H.263+ encoder/decoder software [1], Using the above algorithms, our H.263/H.263+ baseline video-encoder implem entation can encode more than 15 fps in QCIF resolution on a Pentium MMX 20 0-MHz computer.