3D Model Alignment for Accurate Street View Feature Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for generating visual feature maps for autonomous driving require costly and effort-intensive vehicle-based multi-sensor surveying systems and suffer from GPS location inaccuracies in street view images, limiting the utilization of previously captured images.
Innovation Solution
A method and system that aligns three-dimensional models with street view images using absolute coordinate positions to render depth maps, extract feature points, and generate visual feature maps, utilizing aerial surveying data and street view images to reduce costs and improve accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If vehicle-based multi-sensor surveying system is used to acquire mapping data, then three-dimensional geometric information and image information can be obtained, but the cost and effort increase significantly
Solution Approach 1:
The patent uses a three-dimensional model as an intermediary to bridge the gap between existing street view images and the required precise geometric information. The three-dimensional model provides the geometric framework that allows accurate projection and alignment of images without requiring expensive multi-sensor systems on every image capture vehicle.
Solution Approach 2:
The patent performs preliminary creation of a three-dimensional model with accurate geometric information before processing street view images. This pre-computed geometric framework enables subsequent image processing to achieve high measurement precision without requiring complex real-time surveying systems.
2Measurement precision
If GPS equipment is mounted on vehicle to obtain high-precision location information for street view images, then location accuracy improves, but cost increases
Solution Approach 1:
The patent creates a virtual copy of the three-dimensional geometric environment through a three-dimensional model. This digital twin allows the system to project and align images using pre-computed geometric data, eliminating the need for expensive high-precision GPS equipment on every image capture vehicle while maintaining location accuracy.
Solution Approach 2:
The three-dimensional model serves as an intermediary that provides accurate geometric reference data. Instead of relying on expensive real-time GPS measurements for each image, the system uses the three-dimensional model as a mediating framework to achieve precise spatial alignment at lower cost.
3Productivity
If previously captured street view images are utilized, then data acquisition cost is reduced, but location information accuracy deteriorates due to GPS errors
Solution Approach 1:
The patent applies feedback by using the three-dimensional model to correct and refine the location information from previously captured street view images. The three-dimensional model provides a reference framework that allows the system to project images accurately and determine corrected positions, effectively feedback-correcting the original GPS-based location data.
Solution Approach 2:
The three-dimensional model acts as an intermediary that enables the reuse of previously captured images while correcting their location inaccuracies. It provides the geometric reference needed to project and align old images with current accurate coordinate systems, thereby maintaining both productivity and precision.
4Manufacturing precision
If three-dimensional geometric information is projected onto street view images to generate depth maps, then visual feature map quality improves, but processing time increases
Solution Approach 1:
The patent performs preliminary computation of the three-dimensional model with accurate geometric information before processing images. By pre-computing the geometric framework and using efficient projection algorithms, the system achieves high visual feature map quality while minimizing the time required for actual image processing and depth map generation.
Data Source
AI summary
A method for generating a visual feature map including: receiving a 3-D model for a specific area including 3-D geometric information expressed in absolute coordinate positions; receiving a first street view image captured at a first node within the specific area; rendering a depth map associated with the first street view image by projecting at least a part of the 3-D geometric information onto the first street view image; extracting a first set of feature points from the first street view image; determining, on the basis of the depth map, absolute coordinate position information of at least some of the first set of feature points; and generating a first visual feature map associated with the first street view image by storing, for each feature point in the first set, the absolute coordinate position information and a visual feature descriptor in association with each other.