3D Building Model Update via Region-by-Region Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing three-dimensional (3D) building models generated from aerial imagery or camera-equipped vehicles have limited texture resolution, geometry quality, and are difficult to update in real-time, lacking robust image data analytics for consumer and commercial use cases.
Innovation Solution
A system and method for real-time updating of 3D building models using a computer vision database platform that processes uploaded images to detect physical changes by region-by-region mapping and pixel-by-pixel registration, allowing for texture refinement and obstruction removal, enabling users to upload facade images for near real-time updates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If 3D building models are generated from aerial imagery or camera-equipped vehicles, then coverage area is improved, but texture resolution and geometry quality deteriorate
Solution Approach 1:
The patent segments the building model creation process into multiple stages: initial 3D model generation from aerial imagery, followed by separate facade image processing and registration steps. This allows different resolution requirements to be met at different stages - aerial imagery provides broad coverage while high-resolution facade images refine specific building surfaces.
Solution Approach 2:
The patent combines multiple data sources - aerial imagery for overall building structure and high-resolution facade images for detailed textures. By merging these different resolution levels through image registration and mapping, the system achieves both broad coverage and high local quality.
2Ease of manufacture
If traditional 3D building models are used, then initial model creation is simplified, but real-time updating capability deteriorates
Solution Approach 1:
The patent performs preliminary actions by pre-processing facade images - detecting building features, generating initial mappings, and preparing registration data in advance. This preliminary work enables rapid real-time updates when new images are provided, as the system already has templates and parameters ready.
Solution Approach 2:
The patent implements a dynamic update mechanism where the 3D building model can be continuously refined by incorporating new facade images. The system dynamically adjusts the model by comparing new images with the existing model, detecting changes, and updating textures and geometry accordingly.
3Loss of information
If comprehensive image data analytics are implemented, then analytical capability is improved, but system complexity deteriorates
Solution Approach 1:
The patent extracts and isolates specific analytical functions - building detection, change detection, and image registration - as separate processing modules. This extraction allows comprehensive analytics to be performed while managing complexity through modular design, where each module handles a specific analytical task independently.
4Manufacturing precision
If region-by-region mapping and pixel-by-pixel registration are performed, then texture resolution is improved, but processing time deteriorates
Solution Approach 1:
The patent segments the processing task into region-by-region mapping followed by pixel-by-pixel registration only in critical areas. This selective approach maintains high texture resolution where needed while reducing overall processing time by avoiding exhaustive pixel-level analysis across the entire building surface.
Solution Approach 2:
The patent applies partial action by performing detailed pixel-by-pixel registration only on selected regions of interest rather than the entire building facade. This partial processing maintains texture resolution quality for important areas while significantly reducing total processing time.
Data Source
AI summary
A system and method for real-time updating of three-dimensional (3D) building models includes receiving a request to analyze building imagery to detect potential physical changes in or around a first building, receiving the building imagery, the building imagery including one or more images of the building, optionally building a first 3D building model (textured or untextured) based on the building imagery, retrieving, from computer storage, a previously stored version of the first 3D building model, comparing, on a region-by-region basis, the first 3D building model against the previously stored version of the first 3D building model, cataloging in computer storage, based on the comparing, changes to the previously stored version of the first 3D building model, where the changes to the first 3D building model represent physical changes to or around the building occurring since a time of the previous stored version of the 3D building model.