1. Digital video Henning Schulzrinne Columbia University Fall 2003

2. Video sources VGA video (computer video output) – component video – R, G, B + H (horizontal sync), V (vertical sync) Analog video from cameras – NTSC or PAL coded color – composite or component video Digital images: – scanners, copiers and fax machines – digital cameras Images = still pictures Video = motion pictures

3. Video types Bi-level images: black and white – fax, printed output (at pixel level) Gray level (monochrome) images Color (continuous tone) Image typepixels per framebits/pixeluncompressed size fax (200 dpi)1700x220013.75 Mb VGA640x48082.46 Mb XVGA1024x7682418.87 Mb

4. Video compression unlike audio, no physiological model (masking) – except lower color sensitivity than gray level statistical redundancy – background correlation – correlations across an image – nearby pixel correlation – frame correlation (motion compensation) subjective redundancy – impact of different impairments – block artifacts, noise, stair step (“jaggies”), …

5. Bilevel (Fax) coding 1728 pixels per line one-dimensional run-length coding  optimized for early memoryless fax machines Huffman coding 20:1 compression for simple text documents MREAD algorithm uses previous scan line – improves compression to 25:1