Point-Cloud Clip Filter for Occlusion-Free Region Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Point clouds generated by laser scanners from multiple viewpoints can be difficult for users to interpret due to inclusion of occluded elements and lack of clear surface definitions, making it challenging to select and isolate specific regions of interest for modeling and operations.
Innovation Solution
A method is introduced that allows users to define a clip region in a first viewer, which includes selecting a specific point and specifying a clip volume, and then applies this region to a second viewer to render only the data points within this volume, effectively filtering out unwanted data and providing a clearer, more interpretable image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If point cloud data from multiple scan positions is combined in a single coordinate system, then the completeness of the scanned scene is improved, but the interpretability and clarity of the rendered image deteriorates due to occluded elements from different viewpoints
Solution Approach 1:
The patent divides the point cloud data into multiple view-specific data sets, where each data set contains only points visible from a particular scan position. This segmentation allows the system to maintain completeness across multiple views while ensuring that each individual rendered image contains only non-occluded elements, thereby resolving the contradiction between scene completeness and image interpretability.
2Quantity of substance
If all data points from multiple viewpoints are included in the rendered image, then the completeness of the scene representation is improved, but the accuracy of surface definition deteriorates due to inclusion of occluded elements
Solution Approach 1:
The patent performs preliminary processing to determine visibility of each data point from the scan position before rendering. By pre-calculating which points are visible and which are occluded, the system can accurately define surfaces in the rendered image while maintaining completeness of the overall scene representation across multiple views.
3Area of stationary object
If point cloud data from multiple locations is combined, then the coverage of the scanned scene is improved, but the difficulty of selecting and isolating regions of interest increases
Solution Approach 1:
The patent segments the combined point cloud data into view-specific data sets, where each set corresponds to a particular scan position. This segmentation simplifies region selection by allowing users to work with smaller, more manageable data subsets that are already organized by viewpoint, thereby reducing the difficulty of detecting and measuring regions of interest while maintaining comprehensive scene coverage.
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
A viewer renders a first image from a point cloud of data of a scanned scene accessed from a data store. A region of interest may then be defined. In response, the image is updated to only include data included within the region of interest. Alternatively, the region of interest may be provided to a second viewer, where a second image is rendered or updated to only include data included within the region of interest.