Information Identification System Using Location and Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image analysis methods struggle to accurately identify information from images captured while moving, particularly in environments with minimal feature differences, such as a moving camera imaging a parking lot or road, leading to inaccurate information identification.
Innovation Solution
An information identification system that registers object information in a database, detects location and feature attributes of the image, and matches them to identify objects, using features like thickness, color, texture, and geometrical attributes, and connects with a database to enhance identification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image analysis is performed using conventional feature amount extraction methods, then information can be identified from static images, but identification accuracy deteriorates when images are captured while moving or when feature differences are minimal
Solution Approach 1:
The patent introduces location information as an additional dimension beyond traditional image feature amounts. By combining spatial location data with image features, the system creates a multi-dimensional identification approach that remains effective even when visual features are similar, thereby improving identification accuracy in moving camera scenarios
Solution Approach 2:
The system pre-registers location information and object information in databases before actual identification occurs. This preliminary preparation of reference data enables rapid and accurate matching during real-time moving camera operations, preventing identification failures due to minimal feature differences
2Measurement precision
If only image feature amounts are used for identification, then the identification process is simple, but identification accuracy deteriorates when objects have similar visual features
Solution Approach 1:
The identification system is segmented into distinct functional modules: location information acquisition, image feature amount extraction, database matching, and object information identification. This modular segmentation allows the system to incorporate multiple information sources without creating a monolithic complex system, maintaining manageability while improving accuracy
Solution Approach 2:
The database is designed to store multiple types of information (location information, object information, image feature amounts) that serve multiple functions: reference for matching, contextual verification, and supplementary identification criteria. This multi-functionality enables accurate identification without proportionally increasing system complexity
3Productivity
If feature amount extraction is performed on images with minimal color and texture differences, then processing speed is maintained, but identification accuracy deteriorates
Solution Approach 1:
Location information acts as an intermediary that bridges the gap when image features are insufficient for differentiation. By first determining location and then using it to guide feature extraction and matching, the system maintains processing efficiency while improving accuracy through contextual information from the database
Data Source
AI summary
The present invention is to provide an information identification system, a method of identifying information, and a program for identifying information to efficiently identify information by considering information on an object existing in the imaged area when identifying information from an image. The information identification system 10 that identifies information on an object imaged in the image previously registers information on the object existing in a predetermined area, detects the location information of the image and identifies which area the location information belongs to, detects the feature amount of the object imaged in the image, and matches the identified area information and the detected feature amount to the database and identifies information on the object imaged in the image.