Full metadata record
| DC Field | Value | Language |
|---|---|---|
| dc.contributor | Faculty of Engineering | en_US |
| dc.contributor.advisor | Lam, Kin-man (EIE) | - |
| dc.creator | Kong, Minna | - |
| dc.identifier.uri | https://theses.lib.polyu.edu.hk/handle/200/10101 | - |
| dc.language | English | en_US |
| dc.publisher | Hong Kong Polytechnic University | - |
| dc.rights | All rights reserved | en_US |
| dc.title | Age-invariant face recognition for high resolution face images | en_US |
| dcterms.abstract | Face recognition has significant applications in modern society. One of the applications of age-invariant face recognition is that it can be applied to find missing children, and to identify them even after they have grown up. What's more, it can also be utilized for screening a watch list and searching for suspects after many years. The major obstacle for these applications is that no sufficient gallery images with subjects at different ages are available for generating recognition models. Thus, efficient facial features and matching schemes are important for this research. It has significant advanced recent years and many face recognition systems has introduced and deployed. However, there still remain many obstacles on age-invariant face recognition as ageing is a more complex progress. Generally, aging progress can affect skin (i.e. pore texture) and shape texture (i.e. wrinkles). Our project was focused on pore-scale facial features, embedding efficient matching scheme. To improve the matching performance, a locally Affine Transformation Model was introduced and applied in our age-invariant face recognition system. Based on deep features, the pore-facial keypoints was detected in the first stage. Then, by combining feature cost and keypoint location, a lower hall convex optimization problem was built and solved. Finally, the transformed keypoints in the target image can be found. In experiments section, experiment 1 shows the superiority of affine transformation on pose-invariant faces. In experiment 2, different matching scheme were compared. For matching accuracy, RANSAC performs good in High Resolution images, while Affine Transformation Model can find the true optimal matching point and perform better in removing outliers by making use of geographic information. In the last experiment, Affine Transformation Model was applied to match cross-age images. This experiment shows the robust of Affine models in cross-age images matching. | en_US |
| dcterms.extent | 60 pages : color illustrations | en_US |
| dcterms.isPartOf | PolyU Electronic Theses | en_US |
| dcterms.issued | 2019 | en_US |
| dcterms.educationalLevel | M.Sc. | en_US |
| dcterms.educationalLevel | All Master | en_US |
| dcterms.LCSH | Hong Kong Polytechnic University -- Dissertations | en_US |
| dcterms.LCSH | Human face recognition (Computer science) | en_US |
| dcterms.LCSH | Image analysis -- Data processing | en_US |
| dcterms.LCSH | Image processing -- Digital techniques | en_US |
| dcterms.accessRights | restricted access | en_US |
Files in This Item:
| File | Description | Size | Format | |
|---|---|---|---|---|
| 991022270857503411.pdf | For All Users (off-campus access for PolyU Staff & Students only) | 1.28 MB | Adobe PDF | View/Open |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/10101