Adaptive Deblocking Filter for Video Block Distortion

The patent applies deblocking filter selectively based on local boundary characteristics. The determination unit evaluates each block boundary individually using criteria such as gradient magnitude and variance, applying the filter only to boundaries that exhibit actual block distortion. This local quality approach ensures that filter application is optimized for each specific boundary rather than uniformly applied to all boundaries, thereby suppressing block distortion where needed while minimizing unnecessary encoding overhead.

The patent dynamically adjusts the deblocking filter strength parameter based on the transformation block size. For larger transformation blocks, the determination unit increases the filter strength parameter to compensate for the more prominent block distortion that occurs with larger blocks. This parameter change approach allows the system to maintain processing efficiency with larger blocks while adapting the filter characteristics to suppress the enhanced block distortion effects.

The patent employs dynamic adjustment of filter strength based on local image characteristics. The determination unit calculates gradient magnitude and variance for each block boundary, using these dynamic measurements to adaptively set the filter strength. Boundaries with high gradient magnitude (indicating edges) receive weaker filtering to preserve edge sharpness, while boundaries with low gradient magnitude (smooth regions) receive stronger filtering to suppress block distortion. This dynamic approach prevents image quality deterioration by adapting filter strength to local content rather than applying uniform strong filtering.