Digital Image Error Concealment via Edge-Aware Pixel Replacement

The patent divides the image into multiple blocks (e.g., 8x8 or 16x16 pixel blocks) and performs motion estimation independently on each block. This segmentation allows the use of simpler, coarser motion estimation algorithms on smaller units while maintaining overall measurement precision through the accumulation of block-level motion information.

The patent applies different motion estimation strategies to different regions of the image based on their characteristics. Regions with smooth motion are processed with standard block-based estimation, while regions around object boundaries use enhanced techniques that consider temporal information and edge detection to maintain precision despite the coarser block-based approach.

The patent performs preliminary actions by analyzing motion vectors and identifying potential occlusion regions before generating the final interpolated frame. By detecting when an object is moving across the frame boundary and predicting occlusion events in advance, the system can apply special handling techniques to maintain reliability while still benefiting from the speed of motion compensated interpolation.

The patent applies error concealment selectively only to pixels that are identified as erroneous or problematic, rather than processing all pixels uniformly. By first identifying which pixels need correction (those with large motion estimation errors or in occluded regions) and then applying concealment techniques only to those specific pixels, the system improves reliability without the excessive computational cost of processing the entire image.