3D Shape Measurement Using Camera Pose and Image Suitability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current structure from motion (SfM) technologies struggle to accurately measure the absolute size of three-dimensional shapes from images, as they primarily reconstruct relative scales, requiring external references for absolute size determination.
Innovation Solution
A measurement method that selects and processes multiple images with varying camera viewpoints to generate a three-dimensional shape, determines image suitability for measurement, and sets measurement points to calculate object size, incorporating steps for image selection, extraction, generation, determination, display, and measurement, with optional reliability calculation and cursor display for accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If structure from motion technology is used to reconfigure three-dimensional shape from multiple images, then relative scale measurement is achieved, but absolute size measurement cannot be performed without external references
Solution Approach 1:
The system uses the subject itself as the reference for measurement. By detecting feature points on the subject and using the known spatial relationship between camera viewpoints, the system can determine absolute dimensions without requiring external reference objects. The subject's own geometric features serve as the measurement reference.
Solution Approach 2:
The patent transitions from two-dimensional image analysis to three-dimensional spatial reasoning. By reconstructing the three-dimensional shape from multiple two-dimensional images and incorporating camera position information, the system adds the spatial dimension necessary for absolute measurement while maintaining relative scale accuracy.
2Measurement precision
If multiple images are processed to generate three-dimensional shape, then measurement accuracy is improved, but processing time and computational complexity increase
Solution Approach 1:
The system performs preliminary processing by pre-detecting feature points in each image and pre-calculating camera position and orientation parameters before the main measurement computation. This preparation step reduces the computational burden during the three-dimensional reconstruction phase, thereby decreasing overall processing time while maintaining accuracy.
Solution Approach 2:
The measurement process is divided into independent segments: feature point detection in each image, camera parameter calculation, three-dimensional shape reconstruction, and measurement computation. This segmentation allows for parallel processing of different images and operations, significantly reducing total processing time while preserving measurement precision.
3Reliability
If feature points are detected on three-dimensional shape, then measurement reliability is improved, but detection accuracy decreases when feature points lie on flat surfaces
Solution Approach 1:
The system applies different feature detection strategies to different surfaces of the three-dimensional shape. For flat surfaces, it uses edge detection and corner identification methods, while for curved surfaces, it employs curvature-based feature detection. This localized approach ensures high detection accuracy across diverse surface geometries, maintaining measurement reliability.
Solution Approach 2:
The system incorporates feedback mechanisms where detected feature points are validated against the reconstructed three-dimensional shape. If feature points on flat surfaces are detected with low confidence, the system iteratively refines the detection or selects alternative feature points, ensuring high measurement reliability while maintaining precision even on challenging flat surfaces.
Data Source
AI summary
In a measurement method, a processor selects one first image included in a plurality of two-dimensional images of a subject. The processor sets one of the selected first image and one second image extracted from the plurality of two-dimensional images on the basis of the selected first image as a measurement image. The processor extracts at least one third image from the plurality of two-dimensional images. The processor generates a three-dimensional shape of the subject on the basis of a position and a posture of a camera when each image is generated. The processor determines whether or not an image is suitable for measurement with respect to at least one of the first image, the measurement image, and the third image.