Authors

Presenter(s)

Almabrok Essa Essa, Sidike Paheding

Files

Download

Download Project (3.0 MB)

Description

Object detection in aerial imagery has received a great deal of attention in recent years and become one of the most popular research areas in the field of surveillance systems. Issues in aerial imagery, such as low resolution, the presence of noise, complex appearances of objects and more importantly viewpoints variations of objects have made the process of intrusion detection on oil pipeline Right-of-Way (RoW) more challenge. Thus, a detection system must be able to extract prominent features from an object which has to be distinct and stable under different conditions during the image acquisition process. In this work, we present a novel scheme that automatically detects intrusions such as construction vehicles and equipment on pipeline RoW from aerial imagery. In the first part of the framework, a region-of-interest detector is employed to extract potential regions that may contain objects and to reduce the search region from imagery that are not considered to be a region-of-interest. Next, we develop a rotation-invariant gradient histogram based descriptor for a robust object representation. Since it is built in grayscale space, it is independent of the color changes. In terms of tackling motion blur and noise introduced by sensors or atmospheric effects, a noise reducing kernel is used to compute the gradient of the region, and then histogram of orientated gradient is computed for each key region obtained from the first step of the algorithm. The final descriptor is built by concatenating the magnitude of fast Fourier transform of orientation histograms over all key regions. In the last phase of the framework, a support vector machine with radial basis kernel is used as the classifier to detect objects in an image.

Publication Date

4-9-2015

Project Designation

Graduate Research

Primary Advisor

Vijayan K. Asari

Primary Advisor's Department

Electrical and Computer Engineering

Keywords

Stander Symposium project

Disciplines

Arts and Humanities | Business | Education | Engineering | Life Sciences | Medicine and Health Sciences | Physical Sciences and Mathematics | Social and Behavioral Sciences

Automatic Intrusion Detection on Pipeline Right-of-Way via Aerial Imagery

Share

COinS