Residual Data Encoding Parameters for Efficient Image Compression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video data encoding and decoding systems face inefficiencies in encoding and decoding residual data, particularly in handling escape values, which can lead to suboptimal compression and decompression of video data.
Innovation Solution
A method of encoding and decoding image data that involves generating encoding parameters based on residual data estimates, using techniques such as discrete cosine transform, quantization, and entropy encoding to improve efficiency, and incorporating a feedback loop for lossless or lossy entropy encoding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If escape values are used to encode residual data exceeding significance map capacity, then data representation completeness is improved, but encoding complexity and data length increase
Solution Approach 1:
The patent segments the residual data encoding into multiple components: significance maps for primary encoding, and escape values for excess data. This segmentation allows the system to handle different ranges of residual values with appropriate encoding strategies, maintaining completeness while managing complexity through structured division of the encoding task.
Solution Approach 2:
The patent changes encoding parameters dynamically by selecting between different escape value encoding modes (first mode with prefix-unary-suffix structure, second mode with alternative structure) based on the characteristics of the residual data. This parameter adaptation optimizes the balance between representation completeness and encoding efficiency for different data scenarios.
2Manufacturing precision
If detailed residual data is encoded to maintain image quality, then image quality is improved, but compression ratio deteriorates
Solution Approach 1:
The patent applies different encoding precision to different regions of the data by using significance maps that identify which residual values require detailed encoding versus those that can use escape value representations. This local differentiation maintains image quality where needed while reducing data size in regions where high precision is less critical.
Solution Approach 2:
The patent uses partial action by encoding only the necessary portion of residual data with high precision through significance maps, while using more compact escape value representations for the remaining data. This partial high-precision encoding maintains adequate image quality while avoiding the excessive data size that would result from encoding all residual data at maximum precision.
3Productivity
If encoding parameters are derived from residual data estimates, then encoding efficiency is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary action by deriving encoding parameters from residual data estimates before the main encoding process. This preliminary parameter derivation allows the subsequent encoding to proceed more efficiently with pre-determined parameters, reducing the overall processing time despite the initial estimation step.
Solution Approach 2:
The patent implements feedback by using the derived encoding parameters to guide the encoding process, where the parameters are continuously refined based on the actual residual data characteristics observed during encoding. This feedback mechanism optimizes encoding efficiency while managing processing time through adaptive parameter adjustment.
Data Source
AI summary
A data encoding method comprises generating one or more encoding parameters; communicating the one or more encoding parameters for use by a decoding process; and encoding data items representing image data samples of an image by generating residual data indicative of a difference between an image data sample and a predicted version of that image data sample, the encoding step performing the encoding in dependence upon the one or more encoding parameters generated by the generating step; in which the generating step comprises deriving an estimate of the residual data in respect of at least a subset of the image data samples and generating the one or more encoding parameters in dependence upon the derived estimate.