Vehicle Route Tracking Using High-Confidence Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle tracking systems using captured images from multiple vehicles may erroneously recognize non-stolen vehicles as stolen due to unfavorable imaging environments, leading to incorrect identification.
Innovation Solution
An information processing apparatus and method that acquires feature information, confidence information, and position information from multiple image capturing vehicles, determines similarity with registered vehicles, and outputs travel routes based on high-confidence features to suppress erroneous recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If vehicle tracking is performed using captured images from multiple vehicles, then the ability to track stolen vehicles is improved, but the risk of erroneous recognition of non-stolen vehicles as stolen increases
Solution Approach 1:
The patent changes the parameter of feature confidence by introducing confidence information that quantifies the reliability of extracted vehicle features. The determination unit uses this confidence information to weight different feature information sources, giving higher weight to high-confidence features and lower weight to low-confidence features. This parameter change allows the system to maintain tracking reliability while reducing identification errors by dynamically adjusting the influence of different image qualities.
Solution Approach 2:
The patent applies partial action by selectively using only high-confidence feature information for vehicle identification rather than processing all captured images equally. The determination unit performs determination based primarily on high-confidence features, and only considers low-confidence features when necessary or when combined with other high-confidence features. This selective approach reduces erroneous recognition while maintaining tracking capability.
2Reliability
If feature information from multiple sources is processed to improve tracking reliability, then the robustness of vehicle identification is improved, but the complexity of the determination process increases
Solution Approach 1:
The patent introduces confidence information as a new parameter that simplifies the determination process by providing a quantitative measure of feature reliability. Instead of implementing complex algorithms to assess image quality and feature reliability, the system uses pre-calculated confidence information to automatically weight different feature sources. This parameter-based approach maintains high reliability while avoiding excessive complexity in the determination logic.
3Measurement precision
If confidence information is used to weight feature information, then the accuracy of vehicle identification is improved, but the loss of information from low-confidence features increases
Solution Approach 1:
The patent applies partial action by not completely discarding low-confidence feature information but rather using it in a weighted manner or only in combination with high-confidence features. The determination unit can still utilize low-confidence features when they provide supplementary information or when multiple low-confidence features converge to support an identification. This approach reduces identification precision loss while maintaining high accuracy through selective and weighted usage of all available feature information.
Data Source
AI summary
An information processing apparatus includes a determination unit that determines whether a target vehicle is a similar vehicle on the basis of the features of a registered vehicle and the features of the target vehicle indicated by feature information, and an output unit that outputs a route along which the similar vehicle has traveled on the basis of a plurality of positions at which a plurality of captured images including the similar vehicle were captured and a degree of confidence of the features of the similar vehicle.