Edge Correction Using Luminance Difference Weights

The patent applies different correction strategies to different regions of the image. It detects edge regions and applies strong enhancement only to those areas, while maintaining original characteristics in non-edge regions. The method uses luminance difference thresholds to identify edge regions and applies chrominance correction selectively to those regions, thereby improving edge sharpness without introducing noise and color fringe defects throughout the entire image.

The patent dynamically adjusts correction parameters based on local image characteristics. It calculates luminance differences between adjacent pixels to determine edge strength, and uses this information to modulate the strength of chrominance correction. The method changes correction parameters adaptively - applying strong enhancement where edges are detected and weaker or no correction where the image is smooth, thus resolving the contradiction between sharpness enhancement and noise introduction.

The patent converts the harmful effect of chrominance signals into a beneficial one by using chrominance information to guide luminance correction. It calculates luminance differences and uses these to determine where chrominance correction should be applied, effectively using the potential harm of color fringe defects as a guide to identify true edge locations and apply correction only where needed, thereby transforming the problematic chrominance signals into useful edge detection information.

The patent introduces luminance difference calculation as an intermediary step between raw image data and final correction. Instead of directly applying strong chrominance correction that causes color fringe defects, the method first calculates luminance differences to identify edge regions, then uses this intermediate information to control the strength and location of chrominance correction, thereby mediating between raw image data and corrected output to eliminate harmful color fringe effects.