Video Decoding with Classified Motion Vector Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video coding standards like HEVC and VVC have limitations in optimizing the order of predictive motion vectors, leading to suboptimal encoding and decoding efficiency due to insufficient utilization of encoded block information.
Innovation Solution
A method involving classification of adjacent encoded blocks based on size, position, and vector occurrence to determine candidate vectors, followed by selecting a predictive vector with minimal encoding cost, and adding a category identifier to coding information for efficient decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If variable length coding is used to encode sequence numbers of predictive motion vectors, then smaller sequence numbers require fewer bits, but the traditional list ordering does not fully utilize encoded block information leading to suboptimal compression
Solution Approach 1:
The patent changes the ordering parameter of predictive motion vectors from traditional spatial/temporal proximity to classification categories based on block characteristics. By reordering vectors according to classification results (e.g., same size, adjacent position, high vector occurrence), the system optimizes the sequence number assignment to minimize encoding bits while maintaining decoding simplicity through category identifier transmission.
Solution Approach 2:
The patent performs preliminary classification of predictive motion vectors into categories based on block size, position, and vector occurrence characteristics before encoding. This preliminary organization allows the system to assign smaller sequence numbers to more relevant vectors in advance, improving compression efficiency without increasing decoding complexity since the classification rules are predetermined.
2Productivity
If traditional predictive motion vector ordering is used, then decoding is straightforward, but encoding and decoding efficiency is limited due to insufficient utilization of encoded block information
Solution Approach 1:
The patent applies different ordering strategies to different categories of predictive motion vectors based on their local characteristics. Vectors in the same category (e.g., same block size, adjacent position, high occurrence) are grouped together and ordered optimally for that specific group, rather than using a single global ordering rule. This localized optimization improves overall encoding efficiency by tailoring the sequence assignment to each vector's characteristics.
Solution Approach 2:
The patent segments the predictive motion vector list into multiple categories based on block characteristics (size, position, vector occurrence). Each category is then independently ordered and assigned sequence numbers. This segmentation allows the system to exploit different statistical properties of different vector groups, improving overall compression efficiency by treating heterogeneous vectors differently rather than uniformly.
3Reliability
If all predictive motion vectors are included in the candidate list, then comprehensive coverage is achieved, but redundant vectors increase encoding cost and reduce efficiency
Solution Approach 1:
The patent extracts and removes redundant predictive motion vectors from the candidate list by identifying and eliminating duplicates based on classification criteria. Vectors with identical characteristics (size, position, occurrence patterns) are detected and reduced to single representatives, maintaining prediction accuracy while significantly reducing the number of vectors that need to be encoded and transmitted.
Solution Approach 2:
The patent applies a threshold-based approach where only predictive motion vectors meeting certain occurrence frequency criteria are included in the final candidate list. By setting a minimum occurrence threshold, the system filters out rare and potentially redundant vectors, keeping only the most relevant ones. This partial inclusion strategy maintains reliability for common cases while improving encoding efficiency by excluding unnecessary vectors.
Data Source
Figure 1~2
Figure 3~4A
Figure 4B~4C
AI summary
Provided in embodiments of the present application are video decoding and encoding methods and apparatus, a computer-readable medium and an electronic device. The video decoding method comprises: according to information of a decoded block adjacent to a block to be decoded, categorizing the decoded block, obtaining one or more categories; on the basis of encoding information corresponding to the block to be decoded, acquiring a category tag corresponding to the block to be decoded; according to the category tag corresponding to the block to be decoded, determining a target category corresponding to the block to be decoded from among the one or more categories; on the basis of a decoded block vector included in the target category, determining a predicted vector of the block to be decoded, and performing decoding processing of the block to be decoded on the basis of said predicted vector.