Adaptive Video Filtering for Block Effect Reduction

The patent applies different filtering intensities to different regions of the image based on local characteristics. Smooth regions with block effects receive stronger filtering, while textural regions preserve their high-frequency details. This is achieved by calculating a filtering strength map that varies spatially across the image, allowing localized adaptation of filtering parameters to preserve important image features while reducing block artifacts where appropriate.

The patent dynamically adjusts filtering parameters based on local image characteristics. By analyzing the variance or gradient in different regions, the system adapts the filtering strength parameter to match the local content type. Smooth regions receive parameters optimized for block effect reduction, while textural regions receive parameters that preserve high-frequency information, thus resolving the contradiction through parameter adaptation.

The patent identifies textural regions through local analysis and applies filtering selectively. By detecting regions with high frequency content or texture patterns, the system applies weaker filtering to preserve details while still applying moderate filtering to reduce block effects. This localized differentiation allows the system to address block effects in textural regions without sacrificing important image details.

The patent accepts the quantifying noise as an inevitable byproduct of compression but converts this harmful effect into a manageable issue through post-processing filtering. Rather than trying to prevent the noise during encoding, the system applies intelligent filtering in the decoded domain that targets block effects while preserving details. This approach allows high compression ratios to be achieved while mitigating the visual impact of quantifying noise through adaptive filtering.