Home Security System.
Loading...
Date
2018
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
University of Management and Technology
Abstract
The study proposes effective and time efficient face recognition system. The proposed method uses a hybrid
form of approach which uses Haar- like features. These features are extracted by applying Voila Jones
algorithm. Subsequently Local Binary Pattern Histograms is used to train on these features. The data used
for experimentation is obtained from Yale University. It includes normal as well as face that are difficult to
identify. The difficulty is due to varying amount of light and variation in poses. The distinguishing feature of
proposed approach in comparison to conventional approaches is found to be stable and highly accurate. On
top of it, it increases the number of images (frames) processed per second. This frame rate, is observed to be
40 which is quite high when compared with the conventional approaches.
Description
Supervised by: Dr. Masroor Ahmad
Keywords
Conventional approaches, Distinguishing feature, Local Binary Pattern Histograms, Voila Jones algorithm