東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Facial Recognition System with Lensf...

Fang, Yi-Chun.

Linked to FindBook

Google Book

Amazon

博客來

Facial Recognition System with Lensfree Encrypted Optics and Neural Network = = 無透鏡加密光學結合神經網路之人臉辨識.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Facial Recognition System with Lensfree Encrypted Optics and Neural Network =/
Reminder of title:	無透鏡加密光學結合神經網路之人臉辨識.
Author:	Fang, Yi-Chun.
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2021,
Description:	56 p.
Notes:	Source: Masters Abstracts International, Volume: 84-02.
Contained By:	Masters Abstracts International84-02.
Subject:	Computer science. -
Online resource:	https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29404197
ISBN:	9798841513766

Facial Recognition System with Lensfree Encrypted Optics and Neural Network = = 無透鏡加密光學結合神經網路之人臉辨識.
Fang, Yi-Chun.

Facial Recognition System with Lensfree Encrypted Optics and Neural Network =無透鏡加密光學結合神經網路之人臉辨識. - Ann Arbor : ProQuest Dissertations & Theses, 2021 - 56 p.

Source: Masters Abstracts International, Volume: 84-02.

Thesis (M.S.)--National Yang Ming Chiao Tung University, 2021.

.

We proposed a face recognition system based on optical encryption, which can prevent privacy concern where the client's face was disclosed directly in the process. We applied a coded mask as for the optical encryption. A sensor in the lensless imaging system captured blurry but still informative intermediate images. The decryption system combined with deep learning-based classifier to learn the face features from the blurry encrypted images successfully. The whole process does not contain any original face images, so it can reduce the risk of personal data leakage. We designed two experiments using neural networks as classifier models. One was a simulation experiment using a monitor to display a face dataset and another utilized actual face for experiment. In the simulation experiment, two different coded masks were tested for performance. Both trained classifiers for respective encrypted images achieve recognition accuracy of 97.5% for testing set, which verified that the deep learning model can successfully learn effective face features from encrypted images. Then we discuss the limitation of working distance in the system and its solutions. Finally, the recognition accuracy reached 94.4% in the actual face experiment with ideal environment. The experiments reveal the potential of our proposed encrypted face recognition system for practical use case. We believe that the system can be continuously improved and will have more widely applications in the future.

ISBN: 9798841513766Subjects--Topical Terms:

523869
Computer science.
Subjects--Index Terms:

Lensless imaging system

Facial Recognition System with Lensfree Encrypted Optics and Neural Network = = 無透鏡加密光學結合神經網路之人臉辨識.
LDR:04128nmm a2200457 4500 001 2403721
005 20241125080158.5
006 m o d
007 cr#unu||||||||
008 251215s2021 ||||||||||||||||| ||eng d
020 $a 9798841513766
035 $a (MiAaPQ)AAI29404197
035 $a AAI29404197
040 $a MiAaPQ $c MiAaPQ
100 1 $a Fang, Yi-Chun. $3 3773995
245 1 0 $a Facial Recognition System with Lensfree Encrypted Optics and Neural Network = $b 無透鏡加密光學結合神經網路之人臉辨識.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2021
300 $a 56 p.
500 $a Source: Masters Abstracts International, Volume: 84-02.
500 $a Advisor: Tai, Ya-Hsiang;Tien, Chung-Hao.
502 $a Thesis (M.S.)--National Yang Ming Chiao Tung University, 2021.
506 $a .
520 $a We proposed a face recognition system based on optical encryption, which can prevent privacy concern where the client's face was disclosed directly in the process. We applied a coded mask as for the optical encryption. A sensor in the lensless imaging system captured blurry but still informative intermediate images. The decryption system combined with deep learning-based classifier to learn the face features from the blurry encrypted images successfully. The whole process does not contain any original face images, so it can reduce the risk of personal data leakage. We designed two experiments using neural networks as classifier models. One was a simulation experiment using a monitor to display a face dataset and another utilized actual face for experiment. In the simulation experiment, two different coded masks were tested for performance. Both trained classifiers for respective encrypted images achieve recognition accuracy of 97.5% for testing set, which verified that the deep learning model can successfully learn effective face features from encrypted images. Then we discuss the limitation of working distance in the system and its solutions. Finally, the recognition accuracy reached 94.4% in the actual face experiment with ideal environment. The experiments reveal the potential of our proposed encrypted face recognition system for practical use case. We believe that the system can be continuously improved and will have more widely applications in the future.
520 $a 本篇論文提出一種採用光學加密的人臉辨識系統，在人臉辨識的過程中，可以有效避免直接擷取清晰人臉所造成的相關隱私問題。我們利用編碼光罩作為光學加密的光學元件，透過此無透鏡成像系統中的感測器所擷取模糊但仍保有資訊的中繼影像，並結合深度學習分類器對中繼影像進行學習，最後得以成功辨識資料庫中的人臉影像。整個取樣過程不包含任何清晰的人臉影像，因此可以降低流出個資的風險。我們設計了兩種使用基於神經網路作為分類器模型的實驗，分別為使用顯示器播放人臉資料庫的仿真實驗與人臉實拍實驗。仿真實驗中，測試兩種不同編碼光罩加密的模糊影像資料庫，兩個加密影像的分類器對於其測試集都有97.5%的正確率，驗證了深度學習模型可以成功對編碼影像成功學習被加密的人臉特徵。接著探討了系統的工作距離限制與解決方案。最後，人臉實拍實驗中，在較理想的環境下可以達到94.4%的正確率。在實驗中驗證了我們的加密人臉辨識系統具有實用的潛力，相信在未來我們所提出的系統可以持續改進並擁有更廣泛的應用。.
590 $a School code: 2648.
650 4 $a Computer science. $3 523869
650 4 $a Information science. $3 554358
650 4 $a Optics. $3 517925
653 $a Lensless imaging system
653 $a Coded mask
653 $a Facial recognition
653 $a Deep learning method
653 $a 深度學習
653 $a 人臉辨識
653 $a 編碼光罩
653 $a 無透鏡成像系統
690 $a 0984
690 $a 0800
690 $a 0723
690 $a 0752
710 2 $a National Yang Ming Chiao Tung University. $b College of Electrical and Computer Engineering Institute of Electro-Optical Engineering. $3 3773996
773 0 $t Masters Abstracts International $g 84-02.
790 $a 2648
791 $a M.S.
792 $a 2021
793 $a Chinese
856 4 0 $u https://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=29404197