Intra Prediction Line Buffer Optimization for Video Decoding

The patent divides the image into multiple blocks and processes them independently. Each block uses a dedicated line buffer for intra prediction, avoiding the need for multiple global line buffers. This segmentation approach maintains prediction accuracy for each block while reducing overall hardware resource requirements.

The patent applies different prediction methods and buffer configurations to different blocks based on their specific characteristics and positions. Each block can use a single line buffer configured according to its local requirements, rather than using multiple line buffers for the entire image, thus reducing hardware complexity while maintaining local prediction accuracy.

The patent segments the prediction process into block-level operations, where each block uses its own line buffer. This eliminates the need for multiple global line buffers and reduces memory bandwidth requirements by processing blocks independently with smaller, more manageable buffer sizes.

The patent uses a single line buffer per block rather than multiple line buffers for the entire image. This partial approach (using buffers only where needed for each block) achieves sufficient prediction accuracy without the excessive memory bandwidth consumption that would result from using multiple global line buffers.