Construct face photo of unknown person by partial information

• Main Author: Qin, Minhui
• Other Authors: Jiang Xudong
• Format: Final Year Project
• Language: English
• Published: 2016
• Subjects: DRNTU::Engineering
• Online Access: http://hdl.handle.net/10356/69359
• Institution: Nanyang Technological University

The objective of the project is to attain face image reconstruction from an occluded unknown face image using only partial components of that person based on principal component analysis (PCA). AR2 database is used to forming both training dataset and testing dataset. And all face images will be processed in computer using MATLAB (MATrix LABoratory). The unknown person may wear sunglasses that cover his or her eyes or wear a mask which covers almost half of the face. During the 8-month project period, divide-and- conquer method is adopted to achieve the final foal step by step, and MATLAB code is programmed according to the PCA algorithm at different stage of the project. A user interface has been successfully designed to increase the interactivity of the project for testing. The GUI makes it more convenient for user to see and compare the original and reconstructed face image. What’s more, it allows user to choose different training dataset and modify the different value of each parameter according to the user’s preference to the features of specific photo group.