Tile-Based Super-Resolution Algorithm Selection for Power Optimization

The patent applies different super-resolution algorithms to different tiles within an image based on their specific characteristics. Each tile is evaluated independently using parameters such as motion vector magnitude, depth range, and primitive count, and the most suitable algorithm is selected for each tile. This allows high image quality to be achieved in regions that need it while using lower power consumption algorithms in regions where it is not necessary, thus resolving the contradiction between image quality enhancement and power usage reduction.

The patent dynamically selects the super-resolution algorithm based on real-time analysis of tile parameters. The system continuously monitors characteristics such as motion vectors, depth variations, and the number of primitives in each tile, and adapts the algorithm selection accordingly. This dynamic approach enables the system to optimize the balance between processing quality and power consumption based on the actual content and requirements of each image region.

The patent divides the image into multiple tiles and applies different super-resolution algorithms to each tile based on its specific characteristics. Tiles with high motion vectors, large depth ranges, or many primitives receive the high-performance algorithm to maintain image quality, while tiles with simpler characteristics use the faster, lower-performance algorithm. This localized approach reduces overall processing time while maintaining quality where needed.

The patent applies the high-performance super-resolution algorithm only partially to the portions of the image that require it, rather than applying it uniformly to the entire image. By identifying and processing only the necessary tiles with complex characteristics using the more computationally intensive algorithm, the system achieves acceptable image quality while significantly reducing total processing time compared to full-image high-performance processing.