Dynamic Reference Vector Pixel Prediction in Image Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image and video coding techniques face limitations in accurately predicting pixel values, particularly due to fixed and static assignment schemes for not yet decoded pixels, which can lead to inefficiencies in prediction accuracy.
Innovation Solution
A dynamic and flexible method is introduced where a reference vector is used to determine a spatial displacement between a current region and a reference region, allowing for accurate prediction of pixel values by overlapping parts of the reference region with the current region, enabling the image decoder to assign pixel values based on a dynamically updated reference vector.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If fixed and static assignment schemes are used for not yet decoded pixels, then the decoding process is simple, but the prediction accuracy deteriorates
Solution Approach 1:
The patent applies dynamics by transitioning from fixed and static assignment schemes to a dynamic reference vector update mechanism. The reference vector is iteratively refined through multiple passes, allowing the prediction process to adapt and improve accuracy while maintaining manageable complexity through structured iteration.
Solution Approach 2:
The patent employs preliminary action by performing initial pixel value assignments using a reference vector before final decoding. This preliminary prediction allows subsequent refinement passes to build upon an already-established baseline, improving overall accuracy without requiring complete reprocessing.
2Measurement precision
If dynamic reference vector update is used, then prediction accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent implements periodic action through multiple refinement passes that iteratively update pixel values using the reference vector. Each pass periodically revisits and refines predictions, with the process structured to converge toward optimal accuracy while controlling computational overhead through a finite number of passes.
Solution Approach 2:
The patent applies partial action by performing reference vector updates on only the necessary pixel regions that require refinement, rather than reprocessing the entire image. This selective approach improves prediction accuracy where needed while maintaining coding efficiency by avoiding unnecessary computations.
3Adaptability or versatility
If overlapping reference region and current region are used, then prediction flexibility is enhanced, but processing complexity increases
Solution Approach 1:
The patent applies the nested doll principle by having the reference region overlap with and contain portions of the current region. This nesting allows pixel values from the reference region to be directly assigned to corresponding positions in the current region, creating a self-contained prediction mechanism that enhances flexibility while managing complexity through spatial reuse.
Data Source
AI summary
A method and arrangement for prediction of pixel values in an image decoder. In an image decoder, a reference vector which is provided by an image encoder is provided 500. An initiation region of pixels is determined 502, which corresponds to a reference region of pixels at the image encoder. The initiation region is spatially displaced in relation to the prediction region according to the reference vector, and a part of the initiation region overlaps a part of the prediction region. Pixel values are assigned 504 to pixels of the prediction region, whose corresponding pixel values in the initiation region are known. Pixel values of the overlapping region of the initiation region are assigned 506 to the corresponding pixels in the prediction region, the pixel values being assigned 504. By determining an overlapping initiation region based on a dynamic reference vector, characteristic variations close to the prediction region are possible to utilise when predicting images, which increases the accuracy of the prediction.