Stereo Imaging Metrology Epipolar Geometry
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current stereo imaging and metrology approaches face limitations in processing speed, accuracy, and the ability to measure targets not within both cameras' fields of view, leading to inefficient real-time processing and variable measurement resolution depending on object distance.
Innovation Solution
A method and system that concurrently acquire image data from multiple cameras to identify shared and exclusive points, calculating their three-dimensional locations, enabling accurate evaluation of objects even when parts are only visible in one camera's field of view, using a computer-implemented process to determine physical locations and evaluate objects in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If template matching is used for stereo imaging and metrology, then measurement accuracy can be achieved, but processing speed deteriorates significantly
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing epipolar lines during a calibration phase before actual measurement. These epipolar lines are computed in advance based on camera geometry and stored for rapid lookup during runtime, eliminating the need for real-time computation of correspondence relationships and enabling fast template matching along constrained epipolar search paths
Solution Approach 2:
The patent segments the image processing task by dividing it into distinct phases: calibration phase for pre-computing epipolar geometry, and measurement phase for rapid template matching along pre-defined epipolar lines. This segmentation allows complex computations to be performed once during calibration, with only simple lookups and comparisons required during actual measurement, thereby improving processing speed
2Area of stationary object
If template matching is performed across the entire image, then complete coverage is achieved, but processing time increases as n2
Solution Approach 1:
The patent introduces epipolar lines as an intermediary constraint that mediates between the need for complete image coverage and the desire for fast processing. By restricting template matching to proceed only along pre-computed epipolar lines rather than across the entire image, the search space is dramatically reduced while still ensuring complete coverage of all visible features through the geometric constraints of stereo vision
Solution Approach 2:
The patent segments the search space from the entire image area down to specific epipolar line segments. During calibration, the full image coverage requirement is translated into a set of discrete epipolar lines and segments. During measurement, template matching is performed only along these segmented paths, reducing processing time from O(n2) to O(n) while maintaining complete coverage through the systematic traversal of all epipolar lines
3Length of stationary object
If stereo imaging is used for distant objects, then measurement range is improved, but measurement accuracy deteriorates due to smaller angular subtension
Solution Approach 1:
The patent replaces the mechanical/optical constraint of fixed camera baseline with a computational approach using epipolar geometry. By substituting the physical limitation of baseline length with mathematical epipolar constraints, the system can accurately measure distant objects where the angular subtension is small. The epipolar lines provide a computational framework that maintains measurement accuracy regardless of object distance, as long as features are visible in both stereo images
4Reliability
If both cameras must see all target aspects, then stereo measurement reliability is improved, but measurement versatility deteriorates
Solution Approach 1:
The patent inverts the conventional requirement by not requiring both cameras to see all target aspects simultaneously. Instead, it uses epipolar geometry to reliably identify and match features that are visible in only one camera's field of view. The epipolar constraints provide a mathematical framework that allows the system to infer three-dimensional positions of features visible in a single image by relating them to features visible in both images through the epipolar geometry
Data Source
AI summary
A solution for evaluating an object using physical three-dimensional locations of the various points on the object derived from image data concurrently acquired by two or more cameras (e.g., stereo image data) is provided. Image data concurrently acquired by at least two cameras at each of multiple instants is processed to identify one or more points of an object visible in the image data. A physical three-dimensional location of each such point can be calculated at each instant using the corresponding image data. Additionally, a physical three-dimensional location of one or more points of the object visible only in the image data acquired by one camera can be calculated for each of the three different instants using the image data in which the corresponding point is visible and the physical three-dimensional location of one or more of the points visible in the image data acquired by at least two cameras.