Face Matching Device Age-Based Reference Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing facial recognition techniques struggle with accurately matching images of individuals across different ages due to changes in appearance over time, leading to decreased recognition accuracy when trying to retrieve or classify images of the same person at various ages.
Innovation Solution
A face matching device that selects reference face images based on the capturing time of both the target and reference images, ensuring that the selected reference images are of individuals whose age is closest to the target image's age, thereby improving matching accuracy by considering the temporal proximity of the image captures.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If facial recognition techniques are used to match registered face images with captured face images, then image retrieval and classification can be performed, but matching accuracy decreases over time due to aging changes in the person's appearance
Solution Approach 1:
The patent segments the reference face images into multiple groups based on capturing time periods. Instead of using a single reference image, the system divides the reference images into first reference images (captured before a reference time) and second reference images (captured at or after the reference time). This segmentation allows the system to select appropriate reference images based on the temporal proximity to the captured image, thereby maintaining high matching accuracy while enabling effective image retrieval and classification across different time periods.
2Measurement precision
If the stored reference image is updated to reflect the most recent appearance of a person, then recognition accuracy for current appearance is improved, but recognition accuracy for historical or future appearances deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-capturing and storing multiple reference face images at different time periods before they are needed for matching. The system stores first reference images captured before a reference time and second reference images captured at or after the reference time. This preliminary collection of reference images across different time periods enables the system to accurately match captured images regardless of when they were taken, maintaining high recognition accuracy for both historical and current appearances without requiring continuous updates.
3Device complexity
If only a single reference face image is stored for each person, then device complexity is reduced, but the ability to handle images of the same person at different ages is compromised
Solution Approach 1:
The patent introduces dynamics by making the reference image selection process adaptive rather than static. The system dynamically selects between first reference images and second reference images based on the capturing time of the target image relative to a reference time. This dynamic selection mechanism allows the system to automatically adjust which reference images are used for matching, thereby enhancing its ability to handle images of the same person at different ages while maintaining relatively simple storage structure.
Data Source
Figure 1
Figure 2~3
Figure 4~5
AI summary
A face matching device (1) of the present invention includes: a target image obtaining section (101) for obtaining a target face image capturing time, that is, the time of capturing a target face image; and a matching use image selecting section (104) for selecting, on the basis of (i) the target face image capturing time obtained by the target image obtaining section (101) and (ii) a reference face image capturing time stored in a reference image database (121), a reference face image for use in matching. With the arrangement, it is possible to carry out accurate matching with respect to a desired face image among a plurality of images including face images which show respective persons of various ages.