Real-time human detection for surveillance applications

Human detection has been an active field of research for years due to its importance in surveillance applications such as video surveillance system for security. Mild accuracy and computational time has been proven for normal methods in features extraction and classification. There is a need to stud...

Full description

Saved in:
Bibliographic Details
Main Author: Hoe, Qi Xiang
Other Authors: Teoh Eam Khwang
Format: Final Year Project
Language:English
Published: 2014
Subjects:
Online Access:http://hdl.handle.net/10356/60173
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-60173
record_format dspace
spelling sg-ntu-dr.10356-601732023-07-07T16:16:06Z Real-time human detection for surveillance applications Hoe, Qi Xiang Teoh Eam Khwang School of Electrical and Electronic Engineering DRNTU::Engineering Human detection has been an active field of research for years due to its importance in surveillance applications such as video surveillance system for security. Mild accuracy and computational time has been proven for normal methods in features extraction and classification. There is a need to study suitable method to improve on the overall performance of the detection system where having additional function such as describing features of the detected human will further enhance the efficiency. The use of Histogram of Oriented Gradient (HOG) is implemented in the detection of human figure in an image where vectors will be obtained from each pixel in the image through the computation of horizontal and vertical components; these vectors will be used to form a histogram for each image. The histograms will be utilized in the Support Vector Machine (SVM) for classification which is used with the function of sliding window in the detection of human figure in the image. Precision in the detection has been enhanced by having the recursive function to detect the human figure in magnification. The overall accuracy has improved with the enhancement of using additional feature of magnitude inclusive for the HOG computation and further enhanced with the aid of additional features of magnitude inclusive and limitation of magnitude for the HOG computation. Furthermore, segmentation has been performed on the head and body of the human figure for the purpose of system integration with the biometrics’ description. A simulation of the real-time application such as recorded video will be used to test on the integrated system. Bachelor of Engineering 2014-05-23T01:39:45Z 2014-05-23T01:39:45Z 2014 2014 Final Year Project (FYP) http://hdl.handle.net/10356/60173 en Nanyang Technological University 102 p. application/pdf
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic DRNTU::Engineering
spellingShingle DRNTU::Engineering
Hoe, Qi Xiang
Real-time human detection for surveillance applications
description Human detection has been an active field of research for years due to its importance in surveillance applications such as video surveillance system for security. Mild accuracy and computational time has been proven for normal methods in features extraction and classification. There is a need to study suitable method to improve on the overall performance of the detection system where having additional function such as describing features of the detected human will further enhance the efficiency. The use of Histogram of Oriented Gradient (HOG) is implemented in the detection of human figure in an image where vectors will be obtained from each pixel in the image through the computation of horizontal and vertical components; these vectors will be used to form a histogram for each image. The histograms will be utilized in the Support Vector Machine (SVM) for classification which is used with the function of sliding window in the detection of human figure in the image. Precision in the detection has been enhanced by having the recursive function to detect the human figure in magnification. The overall accuracy has improved with the enhancement of using additional feature of magnitude inclusive for the HOG computation and further enhanced with the aid of additional features of magnitude inclusive and limitation of magnitude for the HOG computation. Furthermore, segmentation has been performed on the head and body of the human figure for the purpose of system integration with the biometrics’ description. A simulation of the real-time application such as recorded video will be used to test on the integrated system.
author2 Teoh Eam Khwang
author_facet Teoh Eam Khwang
Hoe, Qi Xiang
format Final Year Project
author Hoe, Qi Xiang
author_sort Hoe, Qi Xiang
title Real-time human detection for surveillance applications
title_short Real-time human detection for surveillance applications
title_full Real-time human detection for surveillance applications
title_fullStr Real-time human detection for surveillance applications
title_full_unstemmed Real-time human detection for surveillance applications
title_sort real-time human detection for surveillance applications
publishDate 2014
url http://hdl.handle.net/10356/60173
_version_ 1772828975374008320