Road Surface Management System Using Feature-Based Image Association
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing road surface management systems using GPS-enabled cameras face challenges in accurately extracting and comparing images at the same location due to GPS measurement errors, leading to incorrect associations of images from different spots.
Innovation Solution
A road surface management system that utilizes an on-board camera with GPS function to collect image data at regular intervals, selects images with high feature amounts, calculates their similarity, and associates them based on location information, ensuring accurate extraction and comparison of old and new images at the same spot by using a feature-based method that accounts for variations in camera position, weather, and observing conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If GPS location information is used to extract and compare images at the same spot, then the system can obtain location data for image comparison, but the measurement error of GPS causes inaccurate extraction of images at the same location
Solution Approach 1:
The patent introduces a map image as an intermediary element between the GPS location data and the actual road surface images. The map image serves as a reference framework that combines GPS location information with visual features, allowing for more accurate matching of old and new images by comparing feature points against the map rather than relying solely on GPS coordinates
Solution Approach 2:
The patent replaces the direct mechanical reliance on GPS coordinate matching with a computational approach using image processing and feature point detection. Instead of trusting GPS location data directly to identify corresponding images, the system uses algorithms to detect and match feature points (such as road markings, signs, and geometric patterns) across images and the map, substituting the mechanical GPS measurement system with an optical-computational system
2Reliability
If automatic detection of feature points is performed to integrate coordinates of old and new images, then corresponding points can be obtained without effects of camera position and weather differences, but the method requires detection of a sufficient number of feature points which increases processing complexity
Solution Approach 1:
The patent applies local quality by focusing feature detection and matching on specific local areas of interest in the images, such as road markings, signs, and geometric patterns, rather than attempting to detect features across the entire image. This localized approach reduces the computational burden while maintaining high accuracy for the critical areas where corresponding points are needed
Solution Approach 2:
The patent segments the image processing task into distinct stages: detecting feature points in individual images, extracting map images at GPS locations, matching features between images and the map, and determining corresponding points. This segmentation allows each stage to be optimized independently and reduces the overall complexity by breaking down the complex matching problem into manageable sub-problems
Data Source
AI summary
According to one embodiment, a road surface management system inputs image data of road surface and capturing information of each of the image data and registers them. The image data are collected by repeatedly capturing a same route at predetermined distance. The capturing information includes information of a location and time at time of capturing. The system selects old and new images at a same spot from the registered image data, performs an association process, extracts an original image which is an old image at an arbitrary spot and a target image which is a new image associated with the original image unit based on the result of the association process, and outputs the original image and the target image.