3D Model Fitting Using Surface Normal Vectors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current three-dimensional model fitting techniques require significant computation time to match the position and orientation of a measurement object with its model, especially when handling objects with indeterminate position and orientation in factory automation settings.
Innovation Solution
An information processing apparatus that acquires a captured image of the measurement object, calculates its surface shape, and limits the position and orientation of the model based on this information, reducing the computational complexity by constraining the movable range of the model to three degrees of freedom.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If three-dimensional model fitting is performed using feature portions extracted from captured images, then position and orientation measurement is achieved, but computation time becomes excessively long
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing surface normal vectors for the three-dimensional model before the actual fitting process. This pre-computed geometric information is then used during model fitting to guide the optimization, significantly reducing the computation time required to match the model with the captured image while maintaining measurement precision.
Solution Approach 2:
The patent changes the parameters used in model fitting from general geometric transformations to specifically constrained transformations based on surface normal vectors. By parameterizing the fitting process in terms of surface normal alignments and pre-computed geometric relationships, the optimization space is reduced, leading to faster computation while preserving measurement accuracy.
2Adaptability or versatility
If a wide settable range for model position and orientation is used, then complete search space is covered, but computational complexity increases significantly
Solution Approach 1:
The patent applies local quality by focusing the model fitting process on specific local features and surface normal vectors rather than performing a global search across the entire parameter space. By locally optimizing the alignment of surface normals and using pre-computed geometric information at critical features, the system achieves accurate positioning within a constrained range, reducing overall computational complexity while maintaining adaptability.
Solution Approach 2:
The patent segments the model fitting process into multiple stages: first extracting and matching surface normal vectors at key features, then using these local matches to constrain and guide the overall model positioning. This segmentation of the fitting process into feature-level and model-level optimization steps reduces the computational complexity of the overall task while maintaining comprehensive search capability.
Data Source
AI summary
An information processing apparatus for matching a position and/or orientation of a measurement object with that of a model of the measurement object includes an acquisition unit configured to acquire a captured image of the measurement object, a calculation unit configured to calculate information indicating a surface shape of the measurement object based on the captured image, and a limitation unit configured to limit a position and/or orientation of the model based on the information indicating the surface shape.