Face recognition using Haar Cascades and Eigenfaces

Denis Kotsyuk

Bachelor's thesis

Abstract:

This thesis focuses on one of the most popular robust face detection methods based on the Viola and Jones framework, that is used on frontal face images, and the face recognition algorithm based on the Principal Component Analysis using the Eigenface method. The theory of these methods and the steps for creating an application that implements these methods are explained. The design of a face recognition system is presented, including image acquisition, grayscale conversion, face detection, face extraction, application of Principal Component Analysis, and face recognition. The process of fine-tuning parameters of the face detector to improve detection accuracy is described and the optimal values of these parameters are selected to minimize the influence of the detection algorithm on the performance of the face recognition algorithm. As a result, a desktop application is designed for face recognition that allows to measure the performance of the model. The classification performance of resulting model is assessed with the confusion matrix and studied in relationship with gender and partial facial occlusions such as glasses and beards.

Keywords

face recognition, Viola-Jones framework, Principal Component Analysis, Eigenface method, EmguCV, C#, face detection

Jazyk práce: English

Datum vytvoření / odevzdání či podání práce: 30. 11. 2020

Obhajoba závěrečné práce

• Vedúci: Ing. Petr Hanzlík, Ph.D.

Czech University of Life Sciences Prague

Faculty of Economics and Management

Bachelor programme / odbor:
Systems Engineering and Informatics / Informatics