Image Structure Descriptor for Pixel Block Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques are computationally intensive and often fail to preserve image structure, leading to undesirable artifacts such as reversed or blurred edges, especially in applications with limited computational resources like mobile devices.
Innovation Solution
The method evaluates the structure of a digital image by determining a structure descriptor for pixel blocks, which represents relationships between pixels, allowing for targeted image processing that differentiates between structure and noise, thereby applying efficient de-noising, sharpening, and de-blurring techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If existing image processing techniques analyze large sections of images to account for structure, then image structure can be considered, but computational intensity increases significantly
Solution Approach 1:
The patent divides the image into small pixel blocks (e.g., 3x3 pixels) and processes each block independently. This segmentation allows the system to capture local structure information without analyzing the entire large image section, significantly reducing computational resource requirements while maintaining structure preservation capability.
Solution Approach 2:
The patent applies different processing approaches to different local regions based on their structure characteristics. By evaluating structure descriptors for each pixel block and determining structure values, the system can apply appropriate processing strength to each local area, preserving structure where needed while reducing computation in uniform regions.
2Manufacturing precision
If existing image processing techniques apply processing to improve image quality, then image quality can be enhanced, but image structure is not preserved and artifacts are introduced
Solution Approach 1:
The patent uses structure descriptors and structure values as feedback mechanisms to guide the image processing. The structure value, derived from the structure descriptor, provides information about the local structure characteristics and is used to control the processing applied to each pixel block, ensuring that processing enhances image quality while preserving structural integrity and avoiding artifacts.
Solution Approach 2:
The patent changes the parameter of processing strength based on the structure value. By adjusting the degree of processing applied to each pixel block according to its structure characteristics, the system can enhance image quality in regions where it improves the image while preserving structure, and reduce or skip processing in regions where it would introduce artifacts.
3Reliability
If existing image processing techniques analyze large sections of images, then structure information can be captured, but processing time increases
Solution Approach 1:
The patent segments the image into small pixel blocks that can be processed independently and in parallel. This segmentation dramatically reduces the processing time compared to analyzing large image sections, as each small block can be processed quickly and the results can be combined to produce the final processed image.
Solution Approach 2:
The patent applies processing to each pixel block individually, which is a partial action compared to processing the entire image at once. This partial processing approach maintains structure information capture capability while significantly reducing the overall processing time through localized operations.
Data Source
AI summary
A structure descriptor for an m×n pixel block of an image may be determined. The m×n pixel block may contain a primary pixel having a primary pixel value and a plurality of secondary pixels having respective secondary pixel values. The structure descriptor may include a plurality of structure indicators each associated with a respective secondary pixel. The respective structure indicators may be based on the primary pixel value and the respective secondary pixel value of the associated secondary pixel. Based on the structure descriptor, a structure value for the m×n pixel block may be determined. Based on the structure value, image processing may be applied to the m×n pixel block.