Image Decoding Error Resistance via Fixed Merging Candidates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image coding and decoding methods using inter prediction lack sufficient error resistance, particularly when information on reference pictures is lost during transmission, leading to discrepancies in bitstream decoding.
Innovation Solution
A method that decodes images on a block-by-block basis by obtaining a fixed number of merging candidates, including prediction directions, motion vectors, and reference picture indices, with empty entries in the merging block candidate list filled with predetermined values to ensure consistent decoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If inter prediction is used for image coding, then compression efficiency is improved, but error resistance deteriorates when reference picture information is lost during transmission
Solution Approach 1:
The patent prepares multiple candidate blocks (first candidate from spatial neighboring block, second candidate from temporal neighboring block, third candidate as alternative) in advance before decoding the current block. This preliminary preparation ensures that when reference picture information is lost, the decoder already has backup candidates ready to use, preventing decoding errors without affecting the compression efficiency of the inter prediction method
Solution Approach 2:
The patent creates a cushioning mechanism by maintaining multiple candidate blocks with different characteristics (spatial, temporal, and alternative candidates). These candidates act as a buffer against potential loss of reference picture information, ensuring continuous reliable decoding while preserving the compression benefits of inter prediction
2Measurement precision
If merging candidates are derived from neighboring blocks, then decoding accuracy is improved, but vulnerability to reference picture loss increases
Solution Approach 1:
The patent applies different qualities/sources to different candidate blocks: the first candidate uses spatial neighboring block information (local spatial context), the second candidate uses temporal neighboring block information (local temporal context), and the third candidate provides alternative information. This diverse local quality approach ensures that if one source is lost, others remain valid, maintaining decoding accuracy while reducing vulnerability
Solution Approach 2:
The patent creates a composite candidate list combining multiple types of candidate blocks (spatial, temporal, and alternative candidates) with different characteristics. This composite structure provides both high decoding accuracy through multiple options and reduced vulnerability to reference picture loss through diversity, resolving the contradiction between accuracy and reliability
Data Source
AI summary
An image decoding method for decoding, on a block-by-block basis, image data included in a coded bitstream includes: obtaining a fixed number of merging candidates each of which is a candidate set of a prediction direction, a motion vector, and a reference picture index which are to be referenced in decoding of a current block (S303); and obtaining, from the coded bitstream, an index for identifying a merging candidate for the current block (S304), wherein the fixed number of merging candidates include: one or more first candidates each derived based on a prediction direction, a motion vector, and a reference picture index which have been used for decoding a neighboring block spatially or temporally neighboring the current block; and one or more second candidates having a predetermined fixed. The fixed number is greater than or equal to two.