3D Point Cloud Clustering for Complex Structure Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing techniques for detecting the shape of complex structures, such as equipment with intricate piping, often misclassify parts of a structure into multiple clusters or combine different structures into one cluster, leading to inaccurate detection of abnormalities.
Innovation Solution
A processing apparatus and method that classify three-dimensional point group data into clusters based on positional information and then determine whether these clusters correspond to a single structure by analyzing their positional relations, using techniques such as projection cluster generation and association based on spatial alignment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If clustering processing is performed on point group data of complex structures, then the structure can be classified into clusters, but one structure may be incorrectly classified into multiple clusters or different structures may be merged into one cluster
Solution Approach 1:
The patent segments the clustering process into two distinct stages: initial clustering based on spatial proximity, and subsequent association clustering based on structural connectivity. This segmentation allows each stage to focus on specific criteria, preventing premature misclassification while maintaining computational efficiency.
Solution Approach 2:
The patent introduces an intermediary association clustering process that acts as a mediator between initial clustering and final structure identification. This intermediary stage uses connectivity information to adjust and refine cluster assignments, correcting errors from the initial spatial-based clustering.
2Productivity
If simple clustering based on positional information is used, then processing speed is maintained, but clustering accuracy deteriorates for complex structures
Solution Approach 1:
The patent divides the clustering process into two sequential segments: a fast initial clustering phase using spatial proximity for quick grouping, and a refinement phase using association clustering for accuracy. This segmentation maintains overall processing speed while improving clustering accuracy for complex structures.
Solution Approach 2:
The patent performs preliminary spatial-based clustering to create initial groups before applying more computationally intensive association clustering. This preliminary action reduces the search space for the refinement stage, maintaining processing efficiency while enabling accurate handling of complex structures.
3Ease of manufacture
If traditional clustering methods are applied to complex equipment with piping, then processing simplicity is maintained, but abnormality detection accuracy decreases
Solution Approach 1:
The patent introduces an intermediary association clustering process that connects spatial clustering with structure identification. This intermediary uses connectivity information to refine cluster assignments, improving abnormality detection accuracy without significantly complicating the overall processing workflow.
Solution Approach 2:
The patent implements feedback mechanisms where association clustering results are used to refine and correct initial clustering outcomes. This feedback loop continuously improves cluster accuracy based on structural connectivity information, enhancing abnormality detection while maintaining processing simplicity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables accurate detection of abnormalities in complex structures by ensuring that point group data is processed to correctly identify and associate clusters representing a single structure, thereby improving the precision of structural inspections.
Implementation Method 1
three-dimensional point group data acquired based on a reflected light from a structure to be inspected illuminated by light
Data Source
AI summary
A processing apparatus (10) includes classification means (12) for classifying three-dimensional point group data acquired based on a reflected light from a structure to be inspected illuminated by light into clusters, which are units of a shape that corresponds to the structure to be inspected, based on positional information at each point of the data; and cluster association means (13) for determining whether a first cluster and a second cluster included in the classified clusters correspond to one structure to be inspected based on a positional relation between the classified clusters.