3D Semantic Image Generation for Flexible Geographic Viewpoints
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image generation technologies based on geographic information struggle with realistic visualization at specific angles, limited versatility, and difficulty in training artificial neural networks with reproduced data.
Innovation Solution
A method and system that utilize a 3D semantic model and artificial neural networks to generate synthetic images from specified locations and directions, incorporating factors like weather, time, and camera movements, enabling realistic terrain visualization and image editing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If satellite image-based image generation technology is used, then terrain information is provided through satellite images, but visualization at specific angles or viewpoints is difficult to satisfy
Solution Approach 1:
The patent transitions from 2D satellite image visualization to 3D virtual scene construction, enabling visualization from multiple dimensions and angles. By building three-dimensional geographic information models, the system allows users to view terrain from arbitrary viewpoints, not limited to top-down satellite perspectives.
Solution Approach 2:
The patent creates virtual copies of real geographic environments using 3D modeling techniques. Instead of directly using satellite images, the system generates synthetic virtual scenes that replicate real-world geographic features, allowing flexible visualization without being constrained by actual satellite imagery limitations.
2Adaptability or versatility
If three-dimensional model-based image generation technology is used, then visualization is provided, but realistic images are difficult to obtain and visualization is limited to pre-modeled local environment
Solution Approach 1:
The patent merges multiple data sources including satellite images, elevation data, and semantic information to create comprehensive 3D geographic models. By integrating these different types of geographic information, the system achieves both broad visualization scope and high image realism simultaneously.
Solution Approach 2:
The patent utilizes parameter changes in the artificial neural network model to adjust image generation characteristics. By modifying network parameters and training data, the system can generate realistic images across various environments, not limited to pre-modeled areas.
3Loss of information
If Geographic Information System (GIS) is used, then detailed information including satellite information and three-dimensional models is obtained, but realistic representation is lacking and intuitiveness is limited
Solution Approach 1:
The patent replaces traditional GIS mechanical data presentation with AI-based synthetic image generation. Instead of displaying raw geographic data and 3D models that require technical knowledge to interpret, the system uses neural networks to generate intuitive visual representations that are easily understood by end users.
Solution Approach 2:
The patent employs color and visual property adjustments in generated images to enhance intuitive understanding. By optimizing color representation, lighting, and visual characteristics through AI processing, the system makes complex geographic information more perceptible and easier to interpret visually.
4Quantity of substance
If existing geographic information reproduction is used, then data is provided, but utilization in training artificial neural network model is difficult
Solution Approach 1:
The patent performs preliminary processing of geographic information data before neural network training. By pre-processing satellite images, elevation data, and semantic information into standardized formats and features, the system makes geographic data readily usable for AI model training, eliminating the need for extensive data preparation work.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A method for generating a synthetic image includes obtaining first location information and first directional information associated with the first location information, obtaining a three-dimensional (3D) semantic model associated with the first location information and the first directional information, generating first content information representing structural information of objects to be generated in a first synthetic image based on the first location information, the first directional information, and the 3D semantic model, generating the first synthetic image based on the first content information using an artificial neural network model, and outputting the generated first synthetic image.