Image-Based Face Search System Using Facial Feature Similarity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Online dating services face challenges in searching for potential matches based on physical appearance, as text-based search methods are inadequate for describing facial features and prone to inaccuracies, and traditional image-based search techniques focus on finding specific individuals rather than similar-looking ones.
Innovation Solution
An image-based search system that allows users to input a query image and compares it to stored images, taking into account user preferences for facial features, using holistic and feature-based perception modes to determine similarity, and displays results based on these criteria, incorporating both text-based and image-based queries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If text-based search is used to search for potential matches, then non-visual attributes (age, gender, ethnicity, location) can be searched effectively, but searching based on physical appearance and facial features is inadequate and prone to inaccuracies
Solution Approach 1:
The patent replaces text-based search mechanisms with image-based search mechanisms. Instead of using textual descriptions to search for facial features, the system uses actual images as queries to search for similar faces in the database, thereby substituting an inadequate mechanical system (text search) with a more effective one (image search).
Solution Approach 2:
The patent uses image copying by allowing users to upload or select query images that serve as templates for finding similar faces. The system compares the query image against stored images to identify matches, effectively using visual copies rather than textual representations to perform the search.
2Measurement precision
If traditional image-based search is used to find specific individuals, then facial recognition accuracy is improved, but the ability to search for similar-looking individuals is lost
Solution Approach 1:
The patent introduces dynamic search capabilities that allow users to adjust search parameters and preferences to find similar-looking individuals rather than exact matches. The system dynamically adapts between precise facial recognition and broader similarity search based on user needs, making the search function versatile rather than fixed.
Solution Approach 2:
The patent changes the search parameters from exact facial feature matching to similarity-based matching. By adjusting the comparison algorithm to evaluate overall facial similarity rather than precise feature correspondence, the system enables users to find individuals with similar appearances while maintaining the ability to perform precise recognition when needed.
3Ease of operation
If users provide textual descriptions of themselves, then profile creation is simple, but accuracy of physical appearance representation deteriorates due to misrepresentation
Solution Approach 1:
The patent replaces the text-based self-description mechanism with image-based representation. Instead of relying on users to accurately describe their physical appearance in text, the system uses uploaded photos to objectively represent their appearance, substituting the unreliable text-based system with a more reliable visual system.
Solution Approach 2:
The patent uses actual photo copies of users rather than their textual descriptions to represent their physical appearance in the database. This eliminates misrepresentation issues by using visual copies that accurately reflect the user's appearance without relying on their ability or willingness to describe themselves accurately.
Data Source
AI summary
A search includes comparing a query image provided by a user to a plurality of stored images of faces stored in a stored image database, and determining a similarity of the query image to the plurality of stored images. One or more resultant images of faces, selected from among the stored images, are displayed to the user based on the determined similarity of the stored images to the query image provided by the user. The resultant images are displayed based at least in part on one or more facial features.