3D Camera Pose Estimation for Structural Crack Change Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems fail to effectively determine changes in structural phenomena, such as cracks or damage, over time due to the panoramic combination of images without considering temporal changes.
Innovation Solution
An image processing apparatus and method that estimates camera positions and postures, selects phenomenon areas, specifies positions in three-dimensional data, and generates projection images to distinguish phenomena across multiple images, allowing for easy detection of changes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If multiple images are panoramically combined to inspect structural cracks, then the ability to observe crack conditions from multiple observation points is improved, but the ability to determine changes in cracks over time deteriorates
Solution Approach 1:
The patent transitions from two-dimensional panoramic image combination to three-dimensional spatial positioning by estimating camera positions and postures. This allows phenomena (cracks) to be identified and tracked across multiple images based on their three-dimensional coordinates, enabling temporal change detection while maintaining ease of observation.
Solution Approach 2:
The system automatically identifies and tracks phenomena across images by comparing their positions in three-dimensional space. This automated feedback mechanism allows the system to detect changes in cracks over time without requiring manual intervention, resolving the contradiction between ease of operation and measurement precision.
2Device complexity
If only panoramic combination of images is performed, then image processing complexity is reduced, but the ability to detect temporal changes in phenomena deteriorates
Solution Approach 1:
The system performs preliminary estimation of camera positions and postures for each image before conducting phenomenon detection. This preliminary action establishes a three-dimensional reference frame that simplifies subsequent phenomenon tracking and change detection, reducing overall processing complexity while improving detection capability.
Solution Approach 2:
The patent introduces camera position and posture information as an intermediary between images and phenomena. This intermediary allows the system to bridge multiple images in three-dimensional space, enabling temporal change detection without significantly increasing processing complexity.
3Measurement precision
If manual identification of phenomena is required, then measurement precision can be maintained, but inspection time and productivity deteriorate
Solution Approach 1:
The system automatically identifies phenomena (cracks) by analyzing image content and tracking them across multiple images based on their three-dimensional positions. This self-service capability maintains measurement precision while significantly reducing inspection time and improving productivity.
Solution Approach 2:
The patent replaces manual mechanical inspection with automated computational analysis. By using image processing algorithms to detect and track phenomena in three-dimensional space, the system maintains accuracy while eliminating time-consuming manual operations.
Data Source
AI summary
An image processing apparatus includes: a posture estimation unit configured to estimate a position and a posture of a camera when the camera has captured each of at least two images including the same phenomena based on the at least two images and three-dimensional data of a structure; an area selection unit configured to select an area of the phenomenon in each of the at least two images; a three-dimensional position specification unit configured to specify a position of each of the phenomena in a coordinate system of the three-dimensional data; and a projection unit configured to generate a projection image by projecting three-dimensional data including the specified position of each of the phenomena and compositing the phenomena included in the images in such a manner that the phenomena for the images are distinguishable from each other.