Pixel-Wise Intra-Prediction for Digital Image Coding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image and video compression standards, such as H.264/AVC and HEVC, face inefficiencies in predicting pixels far from reference pixels, leading to higher prediction errors and bitrates, especially in noisy images or screen content, due to block-wise prediction methods that are not optimal for all types of content.
Innovation Solution
A pixel-wise prediction method that determines a predicted value based on the highest similarity pixel within a defined template region, reducing complexity and eliminating the need for solving linear systems, and is backward adaptive, allowing for efficient compression without restricting the number of reference pixels and minimizing side information transmission.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If block-wise prediction is used to compress image data, then compression efficiency is improved for some content, but prediction error increases for pixels far from reference pixels and bitrate increases for signaling prediction types
Solution Approach 1:
The patent divides the image into blocks and further segments each block into multiple prediction units, allowing different prediction modes to be applied to different segments. This enables fine-grained adaptation to local image characteristics while maintaining compression efficiency.
Solution Approach 2:
The patent dynamically selects prediction modes and reference pixels based on local image characteristics such as gradient directions and noise levels. The prediction strategy adapts to each block's specific features rather than using a fixed approach, improving prediction accuracy for diverse content types.
2Measurement precision
If block size is reduced to improve prediction accuracy, then prediction error decreases, but number of blocks increases leading to higher bitrate for signaling
Solution Approach 1:
The patent applies full block-wise prediction only where necessary based on image characteristics, and uses simpler prediction methods in regions where they suffice. This partial application of complex prediction reduces signaling overhead while maintaining accuracy where needed.
Solution Approach 2:
The patent changes prediction parameters such as block size, prediction mode, and reference pixel selection based on local image characteristics like gradient magnitude and direction. This adaptive parameter adjustment optimizes the trade-off between prediction accuracy and signaling bitrate for different image regions.
3Measurement precision
If template-based prediction with weighted sum is used, then prediction accuracy improves, but computational complexity increases due to solving linear systems
Solution Approach 1:
The patent uses simple averaging and gradient-based prediction methods that require minimal computation, replacing complex linear system solutions. These simpler prediction approaches are computationally inexpensive and sufficient for many image regions, reducing overall complexity.
Solution Approach 2:
The patent applies different prediction strategies to different image regions based on their characteristics. Simple prediction methods are used in smooth regions while more sophisticated methods are applied only where needed, reducing overall computational complexity while maintaining accuracy where required.
4Ease of operation
If reference pixels with noise are used for prediction, then prediction may be performed, but prediction quality deteriorates due to suboptimal reference pixels
Solution Approach 1:
The patent evaluates reference pixel quality using gradient-based metrics and noise detection, and selectively uses or discards reference pixels based on their quality assessment. This feedback mechanism ensures that only high-quality reference pixels are used for prediction, improving overall prediction quality.
Solution Approach 2:
The patent adjusts prediction parameters such as reference pixel selection, weighting factors, and prediction mode based on detected image characteristics including noise levels and gradient directions. This adaptive parameter adjustment optimizes prediction quality by avoiding noisy reference pixels while maintaining prediction capability.
Data Source
AI summary
In a method for coding a sequence of digital images, a prediction error between predicted values and the original values of pixels is processed for generating the coded sequence of digital images. A preset prediction mode is an intra-prediction mode based on pixels of a single image and includes, for a region of pixels with reconstructed values in the single image and for a template of an image area, comparing a first patch of pixels in the region that surround a first pixel to be predicted based on the template with several second patches. A predicted value of the first pixel is determined based on the values of one or more second pixels that have the highest similarity described by the similarity measure among all second pixels of the plurality of second pixels in the region.