Edge Preserving Smoothing Using Pixel Reliability Adjustment

The patent applies local quality by differentiating the treatment of pixels based on their local characteristics. It calculates reliability values for each pixel in the low-resolution image based on the variation degree of corresponding high-resolution pixels, then adjusts the reflection ratio locally for each pixel. This allows edges with high reliability to be preserved while smoothing regions with low reliability, resolving the contradiction between processing speed and edge preserving performance.

The patent changes parameters by introducing reliability values and reflection ratios that dynamically adjust the contribution of low-resolution image pixels to the smoothing process. Instead of uniformly using the low-resolution image, the system modulates the reflection ratio parameter based on local reliability assessments, enabling adaptive edge preservation that maintains both speed and quality.

The patent applies preliminary action by pre-calculating reliability values for each pixel before performing the main smoothing operation. The reliability assessment is conducted in advance based on the variation degree of corresponding high-resolution pixels, allowing the subsequent smoothing process to efficiently use these pre-computed values without adding significant computational burden during the critical smoothing phase.

The patent segments the image processing into distinct stages: generating the low-resolution image, calculating reliability values for each pixel, determining reflection ratios, and performing the smoothing operation. This segmentation allows each stage to be optimized independently, managing computational complexity by breaking down the complex task into manageable, sequential steps.