Stereo Disparity Correction for Rooftop 3D Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for generating three-dimensional feature data, particularly for rooftop structures, face challenges due to limited resolution in laser data, high labor costs, and limitations in processing tall buildings and densely populated areas, with existing technologies struggling to capture detailed rooftop features accurately and efficiently.
Innovation Solution
A device and method that utilize stereo images to extract and classify lines representing rooftop structures, correct disparities, and cluster line pairs to generate precise three-dimensional models, incorporating a stereo disparity calculating unit, line extracting unit, line classification unit, and plane combining unit to create detailed rooftop models.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If laser scanning is used to generate three-dimensional feature data, then the method is automated and covers large areas, but the resolution is limited and rooftop structures cannot be expressed precisely
Solution Approach 1:
The patent combines laser scanning data with aerial photograph data to create a hybrid three-dimensional modeling system. The laser data provides automated large-area coverage while the aerial photographs provide high-resolution rooftop structure details, merging the advantages of both data sources to resolve the contradiction between automation and precision.
Solution Approach 2:
The patent transitions from relying solely on single-point laser measurements to utilizing two-dimensional aerial photograph images as an additional data dimension. This dimensional enhancement allows the system to capture detailed rooftop structures that laser scanning alone cannot resolve, while maintaining automated processing capabilities.
2Measurement precision
If ground-based image shooting is used to capture building details, then high-resolution images can be obtained, but manual work increases and tall buildings cannot be processed
Solution Approach 1:
Instead of shooting images from the ground upward, the patent inverts the approach by capturing images from the air downward using aerial photography. This inversion allows the system to process tall buildings that are inaccessible from ground level while maintaining automated processing through computer-based stereo image analysis.
Solution Approach 2:
The patent uses aerial photographs as optical copies of building structures, eliminating the need for physical presence at each building site. This copying approach automates the data collection process while capturing sufficient detail for three-dimensional modeling, reducing manual intervention to systematic image processing.
3Ease of manufacture
If conventional line extraction methods are used on stereo images, then processing is simpler, but detailed rooftop structures cannot be captured
Solution Approach 1:
The patent segments the line extraction process into multiple specialized stages: initial line extraction from stereo images, classification of lines by type (building contours, rooftop features, terrain), and selective refinement. This segmentation allows the system to maintain processing simplicity while capturing detailed rooftop structures through targeted analysis of different line categories.
Data Source
AI summary
A stereo disparity calculating unit calculates the predicted value of the stereo disparity. A line extracting unit performs line extraction in an image. A line classification unit classifies the extracted lines into different line types. A meaningless line eliminating unit eliminates lines not existing in the real world away from the following processing. A stereo disparity correcting unit corrects the predicted value of the disparity based on the line pairs determined by the line pair determining unit. A line pair clustering unit clusters all the line pairs belonging to the same feature as one cluster. A plane combining unit finds out the location relationship in the three-dimensional space among all the planes of each feature extracted by a plane extracting unit, and generates a three-dimensional model describing the overall structure for each feature.