Multi-Room Image White Balance Using Region-Specific Correction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis techniques struggle to accurately perform white balancing corrections for images of multiple rooms with varying illumination levels, often assuming a single illumination source, leading to inefficiencies in capturing and utilizing building interior information.
Innovation Solution
A system that automatically analyzes and manipulates visual data of images showing multiple rooms, performing room-specific white balancing corrections by identifying neutral pixels and applying chromatic adjustments to correct for different illumination levels in each room, using machine learning and image processing techniques.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing white balancing techniques assume a single illumination source for an entire image, then the processing is simple and fast, but the correction accuracy deteriorates in multi-room images with varying illumination levels
Solution Approach 1:
The patent divides the image into multiple spatial regions corresponding to different rooms or areas with distinct illumination characteristics. Each region is processed independently with its own white balancing correction, allowing accurate handling of varying illumination levels across the entire image while maintaining manageable processing complexity through localized operations.
Solution Approach 2:
The patent applies different white balancing corrections to different spatial regions of the image based on local illumination conditions. By identifying neutral pixels within each region and applying region-specific chromatic adjustments, the system achieves high correction accuracy for each area while the overall process remains systematic and manageable.
2Measurement precision
If existing white balancing techniques are applied to multi-room images, then they are simple to implement, but they fail to correct for different illumination levels in different rooms
Solution Approach 1:
The image is segmented into multiple spatial regions that correspond to different rooms or areas with distinct illumination characteristics. This segmentation enables the system to adapt to varying illumination conditions in each region while maintaining a systematic processing approach that is relatively simple to implement.
Solution Approach 2:
The patent dynamically adjusts white balancing parameters for each spatial region based on local illumination conditions. By identifying neutral pixels in each region and computing region-specific correction factors, the system achieves high adaptability to varying illumination conditions while the overall methodology remains straightforward to implement.
3Loss of information
If floor plans are used to provide building interior information, then layout information is available, but construction and maintenance difficulty increases
Solution Approach 1:
The patent uses captured images as direct copies of the building interior, eliminating the need to manually construct floor plans. The white balancing corrections enhance the quality of these visual copies, providing complete building interior information while avoiding the construction and maintenance difficulties associated with traditional floor plans.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
Techniques are described for using computing devices to perform automated operations involving, for an image acquired at a building and having visual data that shows portions of multiple separate rooms or other spatial regions at the building, automatically analyzing and manipulating visual data of the image to perform white balancing corrections separately for different parts of the image corresponding to the different spatial regions, such as to separately correct for different illumination levels in the different spatial regions. The described techniques may further include using one or more resulting modified images with white balancing corrections in one or more further automated manners, such as to determine one or more types of building information (e.g., a building floor plan) for an as-built multi-room building, and optionally further using the determined building information to provide navigational data for the building.