Image Query System Using Visual Feature Matching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current search engines are inadequate for querying and retrieving non-textual data, such as images, due to their reliance on keyword-based searches, limiting the accessibility and utilization of vast amounts of digitized graphic and multimedia materials.
Innovation Solution
An image query system that allows users to attach implicit and explicit metadata to digital images, enabling querying and retrieval of visually similar images by content and context, using photomatching technology and federated databases to analyze and match image queries with relevant information across multiple repositories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If keyword-based search engines are used to search for images, then the search system is simple to implement, but the ability to query and retrieve non-textual data is inadequate
Solution Approach 1:
The patent introduces an intermediary layer between the user query and the image database. This intermediary consists of feature extraction modules that convert image data into a standardized format, and matching algorithms that compare queries with images. This mediator enables versatile non-textual data querying while managing system complexity through modular design.
Solution Approach 2:
The patent replaces traditional keyword-based mechanical search mechanisms with content-based analysis mechanisms. Instead of relying on textual keywords, the system uses visual feature extraction, pattern recognition, and semantic analysis to match images with queries, enabling direct querying of non-textual data.
2Quantity of substance
If large quantities of digitized materials are stored, then the database capacity increases, but the materials become difficult to query and retrieve
Solution Approach 1:
The patent segments the image database into manageable collections based on visual features, categories, and metadata. Instead of treating the entire database as a monolithic structure, the system divides it into sub-databases organized by visual characteristics, making querying more efficient and intuitive while maintaining large database capacity.
Solution Approach 2:
The patent transforms the querying approach by changing the parameters used for search. Instead of relying on textual keywords, the system uses visual parameters such as color, shape, texture, and spatial relationships. This parameter transformation enables effective querying of large datasets by matching visual features rather than requiring textual descriptions.
3Productivity
If textual keyword searches are used, then the search process is fast and simple, but the accessibility of graphic and multimedia materials is limited
Solution Approach 1:
The patent performs preliminary action by pre-processing and indexing image data before actual queries are received. The system extracts visual features, generates descriptions, and organizes images in advance, so that when queries arrive, the matching process is rapid. This preliminary preparation maintains high search speed while enabling versatile access to graphic materials.
Solution Approach 2:
The patent creates a universal search mechanism that can handle multiple types of queries (textual, visual, semantic) within a single system framework. The search engine can process both traditional keyword searches and content-based queries, making it multi-functional and adaptable to different user needs while maintaining efficiency.
Data Source
AI summary
An image query system, and associated method and computer program product are disclosed that enhance and accelerate image-based querying. To this end, the present invention presents a digital image query with implict and explicit associated metadata and retrieval of visually similar images or other pertinent information from a large collection of images stored on local, networked, or web-linked repositories.