Adaptive De-blocking Filter for MPEG Video Artifacts

The patent applies different filtering strengths to different block boundaries based on local characteristics. The filter adapts its operation by examining activity values and gradient magnitudes at each boundary, applying strong filtering only where blocking artifacts are present while preserving edges elsewhere. This local adaptation resolves the contradiction by maintaining high compression efficiency through selective rather than universal filtering.

The patent implements dynamic filtering where the filter strength and range are adjusted based on local image characteristics. The filtering process is not static but adapts to each block boundary by calculating activity values and gradient magnitudes, dynamically determining whether to apply filtering and at what intensity. This dynamic approach allows the system to maintain coding efficiency while reducing blocking artifacts only where necessary.

The patent applies filtering selectively rather than uniformly across all block boundaries. By calculating activity values and gradient magnitudes, the filter determines partial action only where blocking artifacts are present, avoiding excessive filtering in regions where it is unnecessary. This partial action approach reduces computational complexity while maintaining visual quality by concentrating filtering resources only where needed.

The patent changes filtering parameters dynamically based on local characteristics. The filter strength, range, and activation are adjusted according to activity values and gradient magnitudes calculated for each block boundary. This parameter adaptation allows the system to reduce computational complexity by using minimal filtering where blocking artifacts are absent while applying strong filtering only where necessary to improve visual quality.

The patent implements partial filtering by evaluating activity values and gradient magnitudes to determine which block boundaries require filtering. At low bit rates where blocking artifacts are most prominent, the filter applies processing only to affected boundaries rather than all boundaries, reducing processing time while maintaining visual quality improvement where it matters most.

The patent adapts filtering parameters based on bit rate conditions and local characteristics. By changing the filter activation and strength parameters dynamically, the system optimizes processing time at low bit rates by applying filtering selectively to block boundaries with high activity values and gradient magnitudes, rather than uniformly processing all boundaries.

The patent segments the filtering process into distinct stages: activity value calculation, gradient magnitude calculation, and filter application. Each stage processes specific aspects of the block boundary independently, allowing the complex adaptive filtering to be broken down into manageable computational steps. This segmentation improves filtering effectiveness while managing device complexity by organizing the algorithm into modular, sequential operations.

The patent uses multiple parameters (activity values, gradient magnitudes, threshold comparisons) to control filtering effectiveness. By adapting filter strength and range based on these parameters, the system achieves high filtering effectiveness for blocking artifacts. The parameter-based control manages device complexity by using simple calculations (absolute differences and comparisons) rather than complex algorithms, allowing adaptive behavior with moderate computational requirements.