A Video-based Face Detection and Recognition System in the Compressed DCT Domain

  • 陳 建宇

Student thesis: Master's Thesis

Abstract

Since face detection (FD) and recognition (FR) in video have received significant attention among both academics and practitioners there is a growing demand for the corresponding real time FD and FR algorithms especially for security applications Although current face detection and recognition systems have reached a certain level of efficiency and/or maturity recognition from video remains less efficient and also has problems since the related images and video are usually handled in compressed form which is especially true in indoor or outdoor unconstrained application environments In this work we propose an efficient method of FD and FR in the DCT domain with low complexity computation which has the potential of perming well in the compressed domain The experimental results show that our proposed method has a face recognition rate of 93% The main contribution of this study is providing a possible solution for real-time face recognition performed in the DCT domain This can not only improve the performance of DCT facial feature extraction but can also enhance the face recognition rate in surveillance systems
Date of Award2017 Feb 13
Original languageEnglish
SupervisorMing-Shi Wang (Supervisor)

Cite this

'