3D Model Creation for Small Diameter Objects Using Point Cloud Clustering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Fixed 3D laser scanners struggle to create accurate three-dimensional models of objects with small diameters, such as overhead cables, due to unevenly spaced inter-point distances and incomplete point clouds.
Innovation Solution
The method involves clustering points in the point cloud data, obtaining a curve through the clusters, extracting point clouds on this curve, and connecting them to create a three-dimensional model, using techniques like DBSCAN and RANSAC for clustering and straight-line approximation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If a fixed 3D laser scanner is used to create three-dimensional models, then the measurement range can be covered, but the point cloud becomes coarse at long distances making it difficult to model small-diameter objects
Solution Approach 1:
The patent segments the point cloud data into multiple clusters based on spatial proximity and density characteristics. By dividing the overall point cloud into smaller clusters, the system can apply different processing strategies to different regions, allowing dense clustering in near-field regions and sparse clustering in far-field regions, thus maintaining modelability of small objects across the entire measurement range.
Solution Approach 2:
The patent implements local quality by creating clusters with different density thresholds and inter-point distance criteria for different spatial regions. Near-field clusters use stricter density requirements to capture fine details of small objects, while far-field clusters use more lenient criteria to maintain coverage. This local adaptation of clustering parameters resolves the contradiction between coverage and precision.
2Stability of the object's composition
If points are complemented to form scan lines with a threshold inter-cloud distance, then continuous models can be created, but objects with significantly long inter-point distances cannot be properly modeled
Solution Approach 1:
The patent makes the inter-point distance threshold dynamic by calculating it based on the specific cluster characteristics and object type. Instead of using a fixed threshold, the system adapts the threshold dynamically according to the local point density, cluster size, and inferred object characteristics, allowing the model to accommodate both continuous structures and objects with large gaps.
Solution Approach 2:
The patent changes the inter-point distance parameter based on cluster properties and measurement conditions. By adjusting this critical parameter dynamically, the system can maintain model continuity for dense objects while also capturing sparse objects with large inter-point distances, thus resolving the contradiction between continuity and adaptability.
3Manufacturing precision
If clustering is performed on point cloud data with unevenly spaced points, then small-diameter objects can be identified, but processing complexity increases
Solution Approach 1:
The patent performs preliminary actions by pre-processing the point cloud data to estimate object boundaries, expected densities, and spatial distributions before clustering. This preliminary analysis allows the system to set informed initial parameters for clustering algorithms, reducing the computational complexity of the actual clustering process while maintaining high accuracy for small object detection.
Solution Approach 2:
The clustering algorithm in the patent is designed to be self-adjusting, automatically adapting its parameters based on the input data characteristics. The system self-calibrates clustering thresholds and parameters by analyzing the point cloud's statistical properties, eliminating the need for manual parameter tuning and reducing overall processing complexity while maintaining detection accuracy.
Data Source
AI summary
There is provided an apparatus for creating a three-dimensional model of an object having a small diameter from point cloud data representing three-dimensional coordinates. The apparatus is configured to cluster points included in the point cloud data, obtain a curve passing through points in a cluster for at least one cluster, extract point clouds disposed on a curve coincident with the curve passing through the points, and create a three-dimensional model of the object by connecting the extracted point clouds.