Point Cloud Registration Global Minimum Search
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The Iterative Closest Point (ICP) algorithm often converges to a local minimum registration error when registering multiple 3D point clouds, rather than a global minimum, due to initial position dependencies.
Innovation Solution
An apparatus and method that repeatedly randomize the initialization of source and target point clouds, utilize multi-dimensional trees (KD-trees), and segment point clouds into voxel units to calculate transformation matrices, ensuring that registration results corresponding to a global minimum error are selected among candidates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If the ICP algorithm is used for point cloud registration, then the registration process can be performed efficiently with a clear algorithmic framework, but the registration error converges to a local minimum rather than a global minimum
Solution Approach 1:
The patent applies preliminary action by performing multiple random initializations of the ICP algorithm before selecting the final registration result. Instead of running a single ICP registration, the system executes the algorithm multiple times with different random starting configurations, then chooses the result with the minimum registration error. This preliminary exploration of multiple possible outcomes prevents convergence to poor local minima and ensures better global optimization.
2Measurement precision
If multiple random initializations are performed to search for global minimum, then the registration accuracy is improved, but the computation time and processing complexity increase
Solution Approach 1:
The patent implements feedback by systematically evaluating the registration error from multiple ICP initialization attempts and using this information to select the optimal result. The feedback mechanism compares registration errors across different initializations and terminates the process when the global minimum is identified, balancing computational effort with registration accuracy.
3Productivity
If the registration error is minimized without considering multiple candidates, then the processing is simpler and faster, but the result may not represent the true global minimum
Solution Approach 1:
The patent applies preliminary action by performing multiple random initializations of the ICP algorithm before selecting the final registration result. Instead of running a single ICP registration, the system executes the algorithm multiple times with different random starting configurations, then chooses the result with the minimum registration error. This preliminary exploration of multiple possible outcomes prevents convergence to poor local minima and ensures better global optimization.
Data Source
AI summary
Disclosed herein are an apparatus and method for searching for a global minimum of a point cloud registration error. The apparatus includes memory in which at least one program is recorded and a processor for executing the program. The program performs collecting multiple registration results in which the registration error between a source point cloud and a target point cloud is a local minimum as candidates and selecting the registration result in which the registration error between the source point cloud and the target point cloud is a global minimum, among the candidates. Collecting the multiple registration results may comprise repeatedly randomly initializing the source point cloud and the target point cloud and registering the initialized source point cloud to the initialized target point cloud to thereby search for a registration result in which the registration error therebetween is a local minimum.