Satellite Imagery 3D Object Standardization Without Vector Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D model generation techniques relying on pre-extracted vector data and high-resolution maps face challenges in accurately reflecting dynamic real-world environments and require additional data, making it difficult for organizations to create suitable and user-friendly 3D objects.
Innovation Solution
An apparatus and method using generative adversarial networks (GAN) for preprocessing satellite imagery, followed by vision transformers and convolutional neural networks (CNN) to generate digital surface models and boundary information, then vectorizing these models into standardized 3D meshes, which can be output in predefined formats.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If pre-extracted and classified vector data from LiDAR or high-precision road maps is used to generate 3D models, then the generation process can be performed, but the models cannot accurately reflect dynamic real-world environments or the variety of object types
Solution Approach 1:
The patent uses satellite imagery as a direct visual copy of the real world, replacing indirect vector data representations. This allows the system to capture dynamic real-world environments and diverse object types directly as they appear, rather than relying on pre-extracted data that may be outdated or incomplete
Solution Approach 2:
The system processes satellite imagery that can be continuously updated, allowing the 3D models to reflect current and dynamic real-world conditions. This replaces static pre-extracted vector data with dynamically updatable imagery-based models
2Measurement precision
If high-resolution maps are used to generate 3D models, then detailed information is available, but additional data such as multi-layered geospatial data, satellite imaging conditions, and digital surface models are required which makes it difficult to secure all necessary data
Solution Approach 1:
The patent makes satellite imagery serve multiple functions simultaneously: it provides both the visual appearance information and the geometric/height information needed for 3D model generation. By training the vision transformer on satellite imagery paired with digital surface models, the system extracts both types of information from a single data source, eliminating the need to separately acquire multiple data layers
Solution Approach 2:
The system combines the functions of multiple data sources (satellite imagery, digital surface models, and visual appearance data) into a unified processing pipeline that uses only satellite imagery as input. The vision transformer integrates height prediction and texture extraction in one model, reducing data acquisition complexity
3Manufacturing precision
If traditional methods are used to generate 3D models from satellite imagery, then the process can be performed, but the quality and standardization of the generated objects are insufficient
Solution Approach 1:
The patent applies parameter changes by using a vision transformer that predicts multiple parameters simultaneously (height, boundaries, textures) from satellite imagery. The system transforms the input imagery through learned parameter mappings to generate standardized 3D models with consistent formats, resolving the contradiction between precision and ease of manufacture
Data Source
AI summary
The present disclosure relates to an apparatus and a method for generating standardized 3D objects based on satellite imagery. An apparatus for generating standardized 3D objects based on satellite imagery according to one embodiment of the present disclosure may include a preprocessing unit configured to preprocess the satellite imagery by upscaling the satellite imagery using a generative adversarial network (GAN), and then resizing and dividing the upscaled satellite imagery, a digital surface model generation unit configured to generate a digital surface model for the preprocessed satellite imagery by applying a vision transformer to the preprocessed satellite imagery, an object boundary information generation unit configured to generate boundary information of individual objects included in the preprocessed satellite imagery using a convolutional neural network (CNN)