Change search
ReferencesLink to record
Permanent link

Direct link
Classification and prediction for flying objects based on behavior model
Mid Sweden University, Faculty of Science, Technology and Media, Department of Electronics Design. (with Supervisor Najeem Lawal)
2014 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

This master thesis relates to machine vision system and object classification. The aim of this paper is to classify the flying objects in images in a computer vision system, for example, an eagle, kite or airplane. In this thesis, large amounts of data will be analyzed and a behavior model will be built for each object as important steps towards improving and automating the object classification system. The application of this thesis is to reduce the deaths of golden and bald eagles due to wind blades.

In this thesis work, a new effective method is presented, namely, a stereo vision system, which is applied in feature selection based on this object classification. Several features are primarily extracted, including the flying height, speed, size and degree of changes in the object parameters. 

For image processing and feature extraction, the video acquisition is the first and essential step. Due to the limitation both of equipment and location, the captured videos still do not allow for the collection of sufficient data. For the classification of two objects, a Support Vector Machine (SVM) and Library for Support Vector Machine (LIBSVM) have been employed and implemented in MATLAB. In addition, a preliminary study in relation to the idea of multi-class classification has been conceived and tested by means of an experiment. 

In relation to building a behavior model, the various feature properties and characteristics were beneficial with regards to developing the accuracy and robustness of the final classification and recognition results. The results gathered from these two methods in terms of SVM and LIBSVM are compared and analyzed in order to identify their differences and to determine a better solution. Additionally, the possible future work for this project will be discussed.

Results show that 98% of the flying objects can be currently classified by using OVO SVMs and the OVR SVMs. Based on the results of the classification, 85.82% of the flying objects could be predicted correctly.

Key words: machine vision system, object classification, behavior model, stereo vision system, image processing, feature extraction, SVM, LIBSVM, MATLAB

Place, publisher, year, edition, pages
National Category
Electrical Engineering, Electronic Engineering, Information Engineering
URN: urn:nbn:se:miun:diva-22207OAI: diva2:726632
Available from: 2014-06-23 Created: 2014-06-18 Last updated: 2014-06-23Bibliographically approved

Open Access in DiVA

LIUHUI(7344 kB)798 downloads
File information
File name FULLTEXT02.pdfFile size 7344 kBChecksum SHA-512
Type fulltextMimetype application/pdf

By organisation
Department of Electronics Design
Electrical Engineering, Electronic Engineering, Information Engineering

Search outside of DiVA

GoogleGoogle Scholar
Total: 798 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

Total: 290 hits
ReferencesLink to record
Permanent link

Direct link