3D Model Scene Point Selection via Server-Side Refinement
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating 3D models of indoor spaces are inaccurate and inefficient, particularly when using mobile devices, and fail to provide reliable measurements for applications like furniture manufacturing, due to incomplete sensor data and the need for advanced processing capabilities.
Innovation Solution
A computer-implemented method for selecting refined scene points in a 3D model, which transfers computational tasks from portable devices to a server, allowing for low-cost digitalization and accurate measurement of complex environments by pre-selecting scene points and refining their descriptors using Euclidean geometry and tree-like structures, enabling precise point selection and distance calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If photogrammetry and 3D reconstruction methods are used to generate 3D models from images and depth measurements, then the model represents surface points and features of physical environments, but the existing techniques are inaccurate and inefficient when using mobile devices
Solution Approach 1:
The patent segments the 3D model data into a hierarchical structure with a coarse mesh representation and a point cloud. This segmentation allows mobile devices to efficiently render the coarse mesh while the server processes and provides detailed point cloud data only for regions of interest, resolving the contradiction between measurement precision and processing efficiency by distributing computational load according to spatial importance.
Solution Approach 2:
The system performs preliminary processing by generating a complete 3D model and its coarse mesh representation on the server before transmission to the mobile device. This preliminary action enables the mobile device to quickly display the model structure without performing heavy computational tasks, thereby improving processing efficiency while maintaining measurement precision through subsequent refinement of selected regions.
2Reliability
If computational tasks are performed on portable devices with limited memory and processing speed, then the device can operate independently, but the accuracy and reliability of 3D model generation and measurement deteriorate
Solution Approach 1:
The patent introduces a server as an intermediary between the mobile device and the 3D modeling process. The server performs complex computational tasks including point cloud processing, mesh generation, and refinement, while the mobile device handles user interaction and displays results. This intermediary approach enables high reliability in model generation without requiring complex computational resources on the portable device itself.
Solution Approach 2:
The system creates a simplified copy of the 3D model in the form of a coarse mesh representation that can be efficiently processed and displayed on mobile devices with limited resources. This copying approach allows the device to work with a lightweight representation while the server maintains and processes the full-resolution data, resolving the contradiction between reliability and device complexity.
3Loss of information
If the complexity and size of the generated 3D model increase, then the model provides more detailed information about the physical environment, but the processing speed and memory requirements on portable devices increase
Solution Approach 1:
The patent applies local quality by providing different levels of detail for different regions of the 3D model. The coarse mesh representation provides overall structural information with low energy consumption, while detailed point cloud data is provided only for specific regions of interest when selected by the user. This approach maintains comprehensive environmental information without requiring high processing energy on the mobile device.
Solution Approach 2:
The system performs partial processing by generating the complete 3D model and all detailed information on the server, then transmitting only the essential coarse mesh to the mobile device. When users select specific regions for detailed analysis, the server performs excessive processing by generating refined point cloud data for those specific areas. This partial/excessive action strategy provides comprehensive information while minimizing the processing energy required on portable devices.
Data Source
Figure 1
Figure 2A
Figure 2B
AI summary
The present invention relates to a computer implemented method for selecting scene points in a 3D model. The method comprises the steps of providing scene being a feature point cloud representation. The feature point cloud representation is uploaded and/or stored on a server. In the next step, a 3D model comprising feature points in the first mesh representation is provided, wherein the first mesh representation is converted from a feature point cloud representation. In the next step, the scene of the 3D motor is displayed. The 3D model is provided or is displayed to a user on a client. The user pre-selects a scene point on the client and such a pre-selected scene point is transferred to a server via its descriptor. Based on the descriptor, scene points in pre-selected point neighbourhood is calculated or called and thereby a second mesh representation having higher information value compared to the first match representation is provided. Such a mesh representation is retrieved to the client and displayed to the user, wherein the user is capable to select a further point. Said second mesh representation overlays the first mesh representation. Thereby, a method provides cost effective and more accurate method of selecting points in the 3D model. The particular advantage can be seen in a distance measurement corresponding to a physical environment.