Corner Detection in Rasterized CT Data via Parameterized Edge Operators
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for determining material interfaces in objects, such as those used in computer tomography, struggle to accurately locate intersections where multiple interfaces meet at angles due to smearing effects, leading to imprecise detection of corners and angles, which is crucial for quality assurance in production.
Innovation Solution
A computer-implemented method that generates a gridded representation of the object using measurement data, where parameter-dependent edge finding operators are applied to determine material interfaces by defining specific analysis directions, allowing for precise detection of complex geometries like right-angled corners by considering the object's geometry and imaging system's Point Spread Function.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional edge detection algorithms are used to determine material interfaces, then the method is simple and computationally efficient, but the precision of corner detection deteriorates due to smearing effects
Solution Approach 1:
The algorithm segments the edge detection process into two distinct phases: first detecting candidate edges using a simplified operator, then refining corner positions through a second optimization step. This segmentation allows the system to achieve high precision without requiring the entire algorithm to be computationally intensive.
Solution Approach 2:
The method performs preliminary edge detection using a fast but less precise operator before applying the more accurate corner optimization algorithm. This preliminary action identifies candidate regions that require refined analysis, reducing the computational burden on the precision-critical steps.
2Measurement precision
If the analysis direction is determined from gray value gradients, then the method is adaptive to local image features, but the precision of corner detection deteriorates due to smearing effects
Solution Approach 1:
The analysis direction is determined dynamically through optimization rather than being fixed or solely gradient-based. The algorithm iteratively adjusts the analysis direction to maximize the contrast between materials at the interface, adapting to local features while avoiding the smearing problem inherent in fixed gradient approaches.
Solution Approach 2:
The method changes the parameter used to determine analysis direction from simple gray value gradients to an optimized direction that maximizes material contrast. This parameter change allows the algorithm to achieve sub-pixel precision in corner detection while compensating for smearing effects through iterative optimization.
3Adaptability or versatility
If multiple analysis directions are defined for each measurement point, then the detection of complex geometries improves, but the computational effort increases
Solution Approach 1:
The algorithm segments the measurement points into different categories based on their likelihood of being corners. Only points identified as potential corners undergo multiple analysis directions and iterative optimization, while other points use simpler detection methods. This segmentation maintains high versatility for complex geometries while controlling overall computational effort.
Solution Approach 2:
Instead of applying multiple analysis directions to all measurement points, the method applies the computationally intensive multi-directional analysis only to a subset of points that are likely to be corners. This partial action approach provides sufficient geometric detection capability without the full computational burden of universal multi-directional analysis.
Data Source
Figure 1a~1c
Figure 2a~2c
Figure 3
AI summary
The invention relates to a computer-implemented method for determining material interfaces of an object from at least one measurement of the object, a rasterized representation of the object being produced by means of the measurement, the rasterized representation having a plurality of measurement points, a measurement point having at least one piece of image information, which indicates a value of a measurement variable for the object at the position of the measurement point. The method comprises the determining of a parameterization of the rasterized representation of the object, the parameterization assigning at least one parameter to each of the measurement points of a subset of the measurement points of the representation, and the applying of at least one parameter-dependent edge-detection operator to the measurement points of the rasterized representation, an edge-detection operator being designed to determine the location of at least one material interface in the rasterized representation. In the determination of the location of a material interface from the image information of a measurement point, the edge-detection operator takes into consideration at least the image information of a subset of the measurement points adjacent to the measurement point in the rasterized representation, at least one of the parameters of a measurement point defining an analysis direction within the rasterized representation. When applied to a measurement point, the edge-detection operators are designed to determine at least one material interface along the analysis direction defined by means of the parameter of the measurement point, for a subset of the measurement points an analysis direction being defined in each case by means of at least two of the parameters of a measurement point of the subset, at least one edge-detection operator being applied to the measurement point for each analysis direction, which edge-detection operator determines the location of at least one material interface along the particular analysis direction.