Video Coding Saliency Map Motion Compensation

The patent applies different coding methods to different regions of the video picture based on their saliency levels. High-saliency regions (containing important visual information) are coded with higher quality methods, while low-saliency regions use more aggressive compression. This local differentiation allows increasing overall compression rate while preserving visual quality in critical areas.

The patent changes coding parameters dynamically based on saliency map analysis. By calculating a saliency map and using it to select coding methods, the system adapts compression parameters to the local importance of different picture regions, enabling higher compression rates without sacrificing perceived visual quality.

By identifying and protecting high-saliency regions through the saliency map, the patent ensures that visually important areas maintain quality even under high compression. Less important regions are compressed more aggressively, achieving better overall data transmission efficiency while preserving essential visual information.

The patent performs preliminary saliency map calculation and analysis before the actual coding process. This preliminary action identifies which regions require quality preservation, allowing the coding process to prioritize these areas and achieve better visual quality at higher compression rates.

The patent explicitly avoids uniform coding by using the saliency map to differentiate coding quality across picture regions. High-saliency portions receive high-quality coding while low-saliency portions use more efficient compression, achieving an optimal balance between visual quality and compression rate that uniform coding cannot attain.

Instead of applying high-quality coding uniformly to all regions (excessive action), the patent applies it only where necessary based on saliency analysis (partial action). This selective approach maintains visual quality where it matters most while achieving better compression rates.