Date of Award:

5-2011

Document Type:

Thesis

Degree Name:

Master of Science (MS)

Department:

Civil and Environmental Engineering

Committee Chair(s)

Robert T. Pack

Committee

Robert T. Pack

Committee

James A. Bay

Committee

Xiaojun Qi

Abstract

A new method for assessing the performance of popular image matching algorithms is presented. Specifically, the method assesses the type of images under which each of the algorithms reviewed herein perform to its maximum or highest efficiency. The efficiency is measured in terms of the number of matches founds by the algorithm and the number of type I and type II errors encountered when the algorithm is tested against a specific pair of images. Current comparative studies asses the performance of the algorithms based on the results obtained in different criteria such as speed, sensitivity, occlusion, and others. These studies are an important resource to understand the behavior of the algorithms and their influence on the results obtained. But they do not account for the inherent characteristics of the algorithms that derive the process through which the matching features are evaluated, filtered, and finally selected. Moreover, these methods cannot be used to predict the efficiency or level of accuracy that could be reached by using one algorithm or the other depending on of the type of images. This ability to predict performance becomes handy in situations where time is a limiting factor in a project because it allows one to quickly predict which algorithm will save the most time and resources.

This study addresses the limitations of the existing comparative tools and delivers a generalized criterion to determine beforehand the level of efficiency expected from a matching algorithm given the type of images evaluated. The algorithms and the respective images used within this work are divided into two groups: Feature-Based and Texture-Based. And from this broad classification only three of the most widely used algorithms are assessed: SIFT, SURF, and FAST. The latter is the only one belonging to the feature-based category. Three types of images were evaluated in this study: planar surfaces, cluttered background, and repetitive patterns. For the purpose of matching planar and very “edgy” objects, such as a boat or a building, the feature-based algorithm (FAST) was found to perform with fewer detection errors than the texture-based algorithms. Conversely, when the images evaluated corresponded to cluttered backgrounds or considerably busy scenes, the texture-based detected a larger number of features and matches. The results of each algorithm are evaluated and presented. The number of false matches is manually determined and also presented in the final results. The conclusion and recommendations for feature works in this subject lead towards the improvement of these powerful algorithms to achieve a higher level of efficiency within the scope of its performance.

Checksum

fe48cfe1a019368182e4222117466444

Comments

This work made publicly available electronically on September 29, 2011.

Download

Share

COinS

Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .