Pattern-Based de-correlation for visual-lossless video compression for wireless display applications

The core of many video coding standards is formed by the Discrete Cosine Transform (DCT) for de-correlating spatial video data. When applied to symmetrical spatial data or patterns as found in graphical data, the DCT results in zerovalued sub-bands at predefined positions in the transform domain. The DCT de-correlation is further improved when scan paths are defined that omit the aforementioned zero-valued sub-bands. When deploying these predetermined scan paths on regular video, Human Visual System (HVS) based quantization is applied to force similar sub-bands to zero, such that the remaining energy matches a particular scan path. As a result, visual-lossless compression of video and graphics up to a factor four (2-4) with a PSNR of 40-45 dB is achieved, making this technique suitable for wireless display applications.