Video decoding methods, video decoding equipment, video encoding methods, video encoding equipment

Abstract

According to one implementation scheme, when encoding a block in in-image prediction mode, the block can be divided into two geometrically positioned sub-partitions using its own in-image mode with its available reference pattern. One sub-partition replicates and uses the in-image prediction mode from the original block, and the other sub-partition uses another explicitly or implicitly signaled in-image prediction mode. After geometrically predicting the partitions, the sample values ​​along the partition boundary are adjusted by a blending process with appropriate weights. Geometrically partitioned in-image prediction can be applied to angular in-image prediction mode, or only to negative in-image prediction mode, or only to a specific in-image prediction mode (e.g., mode 34). The choice of transformations or other in-image encoding tools (i.e., in-image sub-partitioning) can be adapted for in-image prediction based on geometric partitioning.