Landmark Position Estimation Using Traffic Signs for Speed Camera Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle navigation systems face challenges in accurately and reliably providing information on the positions of speed cameras due to low image identification, leading to inconsistent and costly maintenance of speed limit enforcement data.
Innovation Solution
A method and apparatus that combines image capturing and machine learning to identify speed cameras by estimating the three-dimensional position of easily identifiable landmarks, such as traffic signs, and then using those positions to accurately locate speed cameras on a virtual plane.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If machine learning through image analysis is used to identify speed cameras, then automation is improved, but measurement precision deteriorates due to low image identification of speed cameras
Solution Approach 1:
The patent introduces traffic signs as intermediary objects that are easier to identify than speed cameras. The system first identifies traffic signs with high confidence using machine learning, then uses these identified traffic signs as reference points to locate speed cameras through spatial relationship analysis on the virtual plane, thereby indirectly achieving speed camera identification with improved precision
Solution Approach 2:
The patent segments the landmark identification process into two distinct stages: first identifying high-visibility landmarks (traffic signs) separately, then using those results to guide the identification of low-visibility landmarks (speed cameras). This segmentation allows each stage to be optimized independently, with the first stage providing reliable reference points for the second stage
2Reliability
If more speed cameras are installed on the road, then reliability of speed limit enforcement information is improved, but device complexity and cost increase
Solution Approach 1:
The patent enables the navigation system to automatically acquire and update speed camera position information through image capturing and machine learning analysis, without requiring manual maintenance or updates. The system self-updates the speed limit enforcement database by processing images and identifying speed cameras autonomously, reducing maintenance complexity and costs
Solution Approach 2:
The patent changes the approach from physical installation of speed cameras to virtual identification through image analysis. By analyzing images captured by the vehicle's camera system and processing them through machine learning algorithms, the system determines speed camera positions without requiring additional physical infrastructure or complex installation procedures
3Productivity
If image capturing and machine learning are combined to identify speed cameras, then productivity is improved, but manufacturing precision deteriorates due to difficulty in detecting speed cameras in images
Solution Approach 1:
The patent performs preliminary identification of traffic signs before attempting to locate speed cameras. By first identifying and establishing the positions of easily detectable traffic signs, the system prepares reference framework that improves subsequent speed camera detection accuracy, making the overall process more efficient without sacrificing precision
Data Source
AI summary
In accordance with an aspect of the present disclosure, there is provided a landmark position estimating method performed by a landmark position estimating apparatus. The method comprises, identifying a first type landmark and a second type landmark from an image, captured by an image capturing device of a vehicle, including various landmarks on a driving route, estimating a three-dimensional position of the identified first type landmark based on a plurality of the images on which the first type landmark is identified and a digital map including a driving area of the vehicle, and estimating a position of the identified second type landmark on a virtual plane including the three-dimensional position of the first type landmark.