Image Decoding Padding for Accurate Boundary Inter Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing demand for high-resolution, high-quality images leads to higher data amounts, increasing transmission and storage costs, and existing inter prediction methods struggle with accuracy when current pictures reference areas outside the boundary of reference pictures.
Innovation Solution
An image encoding/decoding method that pads a motion compensation area and a repetitive padding area around the current picture using motion vectors and pixel values to create an extended picture, improving prediction accuracy through bilateral inter prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the current picture references the outside area of the reference picture boundary to improve prediction accuracy, then the prediction accuracy is improved, but the decoder complexity and memory usage increase
Solution Approach 1:
The patent applies preliminary action by pre-padding the reference picture with motion compensation padding and repetitive padding before the actual prediction process. This prepares the reference picture in advance with valid pixel data at its boundaries, eliminating the need for complex runtime handling of boundary cases during decoding, thus reducing decoder complexity while maintaining prediction accuracy
Solution Approach 2:
The patent introduces motion compensation padding and repetitive padding as intermediary elements between the reference picture and the current picture. These padding areas act as mediators that provide valid reference data for boundary blocks, allowing accurate prediction without requiring the decoder to handle complex boundary conditions dynamically
2Measurement precision
If the current picture references the outside area of the reference picture boundary to improve prediction accuracy, then the prediction accuracy is improved, but the memory usage increases
Solution Approach 1:
The patent applies local quality by differentiating the treatment of different regions in the reference picture. Motion compensation padding is applied specifically to boundary areas where reference data is needed, while repetitive padding is applied to extend this coverage. This localized approach ensures memory is used efficiently only where necessary for prediction accuracy, rather than uniformly across the entire picture
3Measurement precision
If high-resolution, high-quality images are transmitted or stored, then the image quality is improved, but the transmission and storage costs increase
Solution Approach 1:
The patent uses copying by creating motion compensation padding and repetitive padding areas that replicate reference pixel data. Instead of transmitting or storing additional high-resolution data, the decoder generates copies of existing reference picture data through padding operations, maintaining image quality while avoiding proportional increases in data transmission and storage requirements
Data Source
AI summary
An image decoding method may comprise padding a motion compensation padding area within a first padding distance from a boundary of a current picture according to a motion vector of a boundary block adjacent to the boundary of the current picture, padding a repetitive padding area within a second padding area from a boundary of the motion compensation padding area according to a value of a pixel adjacent to the boundary of the motion compensation padding area, and storing an extended picture composed of the current picture, the motion compensation padding area and the repetitive padding area in a memory.