Nokia details context-based predictive video compression
Nokia Bell Labs has published research on a new context-based predictive coder for video compression, offering enhancements for lossless and near-lossless coding efficiency.
Essen – Nokia Oyj's research arm, Nokia Bell Labs, has detailed a novel approach to video compression using a context-based predictive coder. This method is designed for both lossless and near-lossless compression, aiming to improve overall efficiency.
The proposed coder adaptively selects the interframe or intraframe coding mode on a per-pixel basis, utilizing only previously reconstructed samples. This approach eliminates the need to transmit overhead information regarding the coding mode selection to the decoder, thereby reducing data complexity.
Furthermore, the technique facilitates lossless coding for multigeneration video sequences, as the same coding mode information is available during subsequent encoding processes. Nokia Bell Labs indicates that this method simplifies the incorporation of error and context modeling, enabling integration with existing image coders like the JPEG-LS standard.
Experimental results suggest that the new coder, when used with the JPEG-LS standard, achieves significant improvements in compression efficiency. This publication, originating from January 2000, outlines fundamental principles in video compression technology.