Aerial Building Texture Mapping for Automated 3D Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating 3D objects from aerial images are inefficient and require manual processes.
Innovation Solution
A method utilizing artificial intelligence models to automate the generation of 3D objects from aerial imagery, including initial spatial object generation, mapping candidate selection, intermediate texture selection, and final texture generation, using neural networks and algorithms for polygon extraction, feature point matching, and building height calculation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual processes are used for generating 3D objects from aerial images, then the process can be completed with simple tools, but the efficiency and productivity are low
Solution Approach 1:
The patent replaces manual mechanical processing with automated computer-based image processing systems. Specifically, it uses aerial images captured by imaging devices and processes them through computational algorithms to generate 3D objects, substituting human manual operations with automated mechanical and computational systems, thereby dramatically improving productivity
Solution Approach 2:
The patent transforms 2D aerial images into 3D spatial objects by changing dimensional parameters. It extracts building footprints, calculates building heights, and generates 3D coordinates from 2D image data, effectively transforming the dimensional parameters of the data to create three-dimensional representations from two-dimensional sources
2Productivity
If automated AI models are used for 3D object generation, then productivity and efficiency are improved, but the device complexity and computational requirements increase
Solution Approach 1:
The patent divides the complex 3D object generation process into distinct sequential steps: aerial image acquisition, building footprint extraction, feature point matching, spatial coordinate calculation, building height calculation, and 3D object generation. This segmentation allows each step to be processed independently using appropriate algorithms, managing system complexity while maintaining high automation and productivity
Solution Approach 2:
The patent introduces intermediate data structures and processing stages as mediators between the input aerial images and the final 3D objects. It uses building footprint polygons, feature point coordinates, and spatial transformation matrices as intermediary representations, allowing complex transformations to be broken down into manageable steps that reduce overall system complexity
Data Source
AI summary
A method for generating three-dimensional objects includes an initial spatial object generation step, a mapping candidate selection step, and an intermediate texture selection step. An aerial image is a building image captured from the air using an imaging device. In the initial spatial object generation step, a processing device generates an initial spatial object, a 3D object corresponding to the building based on aerial images. A building face is defined on a surface of the building. A building face normal vector perpendicular to the building face is defined. At least one of the aerial images includes an initial texture corresponding to the building face. A photographing direction vector in the direction of the building face from the imaging device is defined for each of the aerial images. The processing device selects, as a mapping candidate aerial image, one of the aerial images that includes the initial texture and for which a dot product between the photographing direction vector and the building-face normal vector is negative. The initial spatial object includes an object surface corresponding to the building face. In the intermediate texture selection step, the processing device selects, as an intermediate texture, the initial texture having the largest area among initial textures corresponding to the mapping candidate aerial images.