Road Map Generation Using Multi-Neural Network Feature Fusion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image processing technologies face challenges in accurately extracting road features from remote sensing images, particularly in terms of road direction and width, leading to inefficiencies in map generation and navigation systems.
Innovation Solution
A method involving multiple neural networks is employed, where a first neural network extracts initial road feature information, and a third neural network, trained with road direction information, enhances this data, which is then fused to generate a more accurate road map, improving feature extraction precision.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional image processing technologies are used to extract road features from remote sensing images, then the processing speed is relatively fast, but the accuracy of road direction and width extraction is insufficient
Solution Approach 1:
The patent divides the road feature extraction task into multiple specialized neural networks: a first neural network for initial feature extraction, a second neural network trained with road width information for width feature extraction, and a third neural network trained with road direction information for direction feature extraction. This segmentation allows each network to specialize in specific aspects, improving overall extraction accuracy while managing complexity through modular design
Solution Approach 2:
The patent creates a multi-functional neural network system where multiple networks work together to perform different aspects of road feature extraction. The first neural network provides general road feature extraction, while the second and third networks add specialized width and direction analysis. This multi-functionality approach enables comprehensive feature extraction that addresses both accuracy requirements and various road attribute measurements
2Measurement precision
If a single neural network is used for road feature extraction, then the system complexity is low, but the extraction precision of road direction and width features is insufficient
Solution Approach 1:
The patent segments the feature extraction function across three distinct neural networks, each trained with specific supervision information. The second neural network is trained with road width information to extract width features, while the third neural network is trained with road direction information to extract direction features. This segmentation enables precise extraction of specific road attributes that a single network cannot achieve
Solution Approach 2:
The patent employs feedback mechanisms where the extracted features from multiple networks are fused together. The first road feature information from the initial network is combined with width features from the second network and direction features from the third network. This feedback loop allows the system to iteratively refine feature extraction accuracy by incorporating specialized information from each network into the final road map generation
Data Source
AI summary
Method and apparatus for generating a road map, electronic device, and non-transitory computer storage medium are disclosed, including: inputting a remote sensing image into a first neural network to extract first road feature information of multiple channels via the first neural network; inputting the first road feature information of multiple channels into a third neural network, to extract third road feature information of multiple channels via the third neural network, where the third neural network is a neural network trained by using road direction information as supervision information; fusing the first road feature information and the third road feature information; and generating a road map according to a fusion result.