3D Building Model Generation from 2D Floor Plans
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in generating realistic 3D models of building interiors and creating indoor navigation networks, particularly for multi-floor buildings, as they require specialized equipment and expertise, and often result in undesirable artifacts.
Innovation Solution
A system that allows non-expert users to generate realistic 3D models and indoor navigation networks by converting 2D floor plans into 3D models using user-input heights and photos, which are then textured and connected across floors to create a navigable network.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If specialized equipment such as Lidar or multi-view camera rigs is used to generate 3D models, then measurement precision and realism of 3D models are improved, but device complexity and cost increase
Solution Approach 1:
The patent uses 2D floor plan images as simplified copies or representations of the actual building spaces. Instead of requiring complex 3D scanning equipment, the system processes readily available 2D architectural drawings to reconstruct 3D models, thereby achieving accurate spatial representation without specialized measurement devices
Solution Approach 2:
The patent replaces mechanical 3D scanning systems (Lidar, camera rigs) with a computational image processing system. The mechanical measurement processes are substituted by algorithmic interpretation of 2D floor plan images, using computer vision and geometric reconstruction techniques to achieve 3D model generation
2Manufacturing precision
If specialized 3D modeling software such as AutoDesk or Blender is used, then manufacturing precision of 3D models is improved, but ease of manufacture deteriorates due to requiring expert knowledge
Solution Approach 1:
The system enables non-expert users to generate 3D models by automatically processing uploaded 2D floor plan images. The software performs self-service functions including automatic geometric reconstruction, texture application, and navigation network generation without requiring user intervention in complex 3D modeling operations, thereby eliminating the need for specialized software expertise
Solution Approach 2:
The system performs preliminary processing of 2D floor plan images to extract geometric information, spatial relationships, and structural features before 3D reconstruction. By pre-processing the input data and automatically preparing it for modeling, the system eliminates the need for users to manually create 3D models from scratch using complex software
3Productivity
If AI is used to convert point clouds into 3D objects, then productivity of 3D model generation is improved, but manufacturing precision deteriorates due to artifacts such as merged objects
Solution Approach 1:
The system performs preliminary processing of 2D floor plan images to extract and separate geometric information for different objects and spaces before 3D reconstruction. By pre-segmenting the input data based on architectural features, walls, and spatial boundaries, the system maintains object separation integrity throughout the generation process, avoiding merged objects artifacts
Solution Approach 2:
Instead of using AI to interpret and reconstruct objects from point cloud data, the system directly copies geometric information from the authoritative 2D floor plan images. This approach preserves the accuracy of object boundaries and spatial relationships as originally defined in the architectural drawings, avoiding the interpretation errors and merging artifacts that occur with AI-based point cloud processing
4Reliability
If manual coding of navigation features is performed after 3D model generation, then reliability of indoor navigation network is improved, but productivity deteriorates due to additional time required
Solution Approach 1:
The patent merges the 3D model generation process with navigation network creation by simultaneously extracting both geometric model data and navigation feature data from the same 2D floor plan images. This integrated approach generates both the visual 3D representation and the functional navigation network in a single processing pass, eliminating the need for separate manual coding of navigation features
Solution Approach 2:
The system performs preliminary extraction of navigation-relevant features (doors, corridors, rooms, spatial relationships) during the initial 3D model generation phase. By identifying and encoding these navigation features while processing the floor plan images, the system prepares all necessary data for both visualization and navigation functionality simultaneously, ensuring reliability without requiring subsequent manual intervention
Data Source
AI summary
A 3D building model generator has a server that constructs a 3D texture model for each floor in a building by extruding a 2D floor plan to 3D using user-inputted heights, and adding photos captured by a smartphone or mobile device as textures to image planes for surfaces in the 3D model. The server displays on the smartphone a floor-plan portion, or a virtual 3D display based on the camera orientation with a facing wall highlighted and confirmed by a smartphone user. The user moves mapping points overlaid upon the photo to wall corners to map the photo to the facing surface in the model. The user flags rooms and exits such as doors and elevators as walkable. The server stacks the 3D texture model for the floors and connects them into a walkable network of the rooms and exits that are flagged as walkable. Hashing removes duplicate floorplans.