Parallel Gain Map Generation Using Overlapping Sub-Images

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing image processing systems face challenges in generating accurate gain maps for arrays of imaging sensors, as different sensors produce varying output values even when viewing the same scene, leading to artifacts when sub-image gain maps are stitched together in parallel processing.

Innovation Solution

The technique involves dividing input images into overlapping sub-images, generating sub-image gain maps, and renormalizing gain values to ensure equal average values in overlapping regions, allowing for the combination of sub-image gain maps into a final gain map without boundary artifacts, facilitating parallel processing and improving accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If multiple sub-image gain maps are generated in parallel from divided input images, then processing speed is improved, but boundary artifacts appear when stitching sub-image gain maps together

Engineering Contradiction:

Improveprocessing speedVSAvoidgain map accuracy

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The input images are divided into multiple sub-images that are processed in parallel to generate sub-image gain maps. This segmentation enables faster processing while the overlapping regions ensure continuity when combining the results into a final gain map.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

Overlapping regions between adjacent sub-images are given special treatment through renormalization. The gain values in overlapping regions are adjusted to ensure continuity and eliminate boundary artifacts, while non-overlapping regions maintain their original gain characteristics.

Inventive Principle:

Principle #3Local quality

2Loss of time

If sub-image gain maps are combined without renormalization, then processing time is reduced, but average gain values differ in overlapping regions causing artifacts

Engineering Contradiction:

Improveprocessing timeVSAvoidgain map consistency

Core Design Contradiction:

Loss of timeVSReliability

Solution Approach 1:

The renormalization of gain values in overlapping regions is performed as a preliminary step before combining sub-image gain maps. This ensures that average gain values are equalized in advance, preventing artifacts and ensuring consistency in the final gain map.

Inventive Principle:

Principle #10Preliminary action

3Productivity

If different imaging sensors are used to capture images, then coverage and parallel processing capability are improved, but different output values are generated even for the same scene

Engineering Contradiction:

Improvesensor array coverageVSAvoidsensor output consistency

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

Gain values are applied as correction parameters to normalize the output of different imaging sensors. By calculating and applying sensor-specific gain factors, the system compensates for variations in sensor responses and achieves consistent output values across the sensor array.

Inventive Principle:

Principle #35Parameter changes

Data Source

PatentUS11563899B2Parallelization technique for gain map generation using overlapping sub-images

Publication Date: 2023.01.24 RAYTHEON CO

AI summary

A method includes obtaining multiple spatially-displaced input images of a scene based on image data captured using multiple imaging sensors. The method also includes dividing each of the input images into multiple overlapping sub-images. The method further includes generating multiple overlapping sub-image gain maps based on the sub-images. In addition, the method includes combining the sub-image gain maps to produce a final gain map identifying relative gains of the imaging sensors. An adjacent and overlapping pair of sub-image gain maps are combined by renormalizing gain values in at least one of the pair of sub-image gain maps so that average gain values in overlapping regions of the pair of sub-image gain maps are equal or substantially equal.