Title

Human tracking with active camera (U-TrAC)

Date of Publication

2006

Document Type

Bachelor's Thesis

Degree Name

Bachelor of Science in Computer Science

Subject Categories

Computer Sciences

College

College of Computer Studies

Department/Unit

Computer Science

Thesis Adviser

Clement Y. Ong

Defense Panel Member


Alexis V. Pantola
Karlo Shape O. Campos
Joel P. Ilao

Abstract/Summary

Tracking of deformable objects is an active area of research in the field of machine vision. It is useful for many vision-based systems such as video surveillance systems. Some difficulties encountered in some of the researchers are the automatic identification of blobs as humans through body shape and motion, avoidance of false alarms due to lighting and background changes, and the robustness of the system to occlusions.

Human tracking with Active Camera (U-TrAC) is a system that detects and tracks human from a given input video stream, taken using an active camera platform. The web camera is placed on an electro-mechanical platform that is positioned at the center of the ceiling of the room to be observed. The system is capable of rotating the camera 360 horizontally and tilting it 90 vertically at a rate of 7 revolutions per minute (rpm) or 40 per second. By panning and tilting the camera around, the system looks for a human body, using a cascaded classifier that is trained with 176 positive and 610 negative samples and consisting of 16 stages for human body detection. The human body moving at a maximum rate of 4.0 km/hr may be detected from 3.3 meters to as much as 16.6 meters away from the camera. The human body could be detected as long as it is standing upright, presenting a front or back view and the contrast ratio of the human body to the background is at most 26:66. The tracking technique is robust to partial occlusions. The system is able to maintain the tracking of the human body in spite of partial occlusions provided the occlusion will not obstruct the centroid or the tracking point of the human body.

Abstract Format

html

Language

English

Format

Print

Accession Number

TU13555

Shelf Location

Archives, The Learning Commons, 12F, Henry Sy Sr. Hall

Physical Description

1 v. (various foliations) : ill. (some col.) ; 28 cm.

Keywords

Image processing--Digital techniques; Computer vision; Cameras; Photography

This document is currently not available here.

Share

COinS