Temporal Filtering Independent Color Channels Image Signal Processor

The patent segments the color processing into independent channels (e.g., luminance and chrominance channels in YCbCr color space). Temporal filtering is selectively applied to specific channels based on their noise characteristics and importance, rather than uniformly to all channels. This segmentation allows the system to reduce noise in critical channels while avoiding unnecessary processing in less sensitive channels, thereby reducing overall processing complexity.

The patent applies different filtering strategies to different color channels based on their local quality requirements. The luminance channel, which carries brightness information critical for perceived image quality, receives stronger temporal filtering. In contrast, chrominance channels carrying color information receive reduced or no filtering, as excessive filtering can cause color artifacts and the human eye is less sensitive to color noise. This localized quality-based approach optimizes the balance between noise reduction and processing efficiency.

The patent implements partial motion compensation by applying it only to specific color channels or specific regions of the image where motion is detected, rather than uniformly to the entire image and all channels. This partial action approach maintains filtering accuracy in areas where motion occurs while reducing computational load in static areas, achieving a balance between precision and energy consumption.

The processing is segmented into motion-detecting regions and static regions, with motion compensation applied selectively. The system divides the image into blocks or regions, identifies those with significant motion, and applies computationally intensive motion-compensated temporal filtering only to those regions. Static regions receive simpler filtering, significantly reducing the overall computational load while maintaining accuracy where needed.

The patent employs periodic action by applying spatial and temporal filtering in alternating stages rather than simultaneously processing all filters. The processing pipeline is organized into periodic passes: one pass applies spatial filtering to remove noise within frames, another pass applies temporal filtering to reduce noise across frames. This periodic alternation allows the system to maintain high image quality through multiple processing stages while managing processing time through efficient staged execution.

The patent applies preliminary spatial filtering to image data before applying temporal filtering. By removing a portion of the noise through spatial filtering in advance, the subsequent temporal filtering operates on already-cleaned data, reducing its computational burden and processing time. This preliminary action approach ensures that multiple processing stages contribute to image quality without linearly increasing total processing time, as each stage builds upon the work of previous stages.