Wavelet-based edge detection

Date of Publication

2004

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

Jocelynn Wong

Defense Panel Member


Clement Y. Ong
Roselle S. Berdin
Joel P. Ilao

Abstract/Summary

There are existing edge detection systems that are robust and effective. The most common algorithms for these systems are the Gradient and Laplacian method. Nonetheless, these algorithms have difficulties in detecting edges when the difference in contrast between the target and the background is low. Another trouble area is that they are quite susceptible to noise.

The Gradient and Laplacian method are implemented in this system on low contrast images with different noise levels. In an attempt to discover a better alternative, the Wavelet algorithm was also applied to the edge detection principle. A comparative study on the speed and accuracy in the detection of edges was then performed on the three algorithms. The speed and accuracy results were quantified through the MATLAB's Stopwatch Timer function and the Pratt's Figure of Merit formula respectively.

The Gradient algorithm had the fastest run-time and the Laplacian algorithm had detected the most number of accurate edges when there is no visual noise in the image. Interestingly, the Wavelet algorithm appeared to be intermediate in both its speed and its speed and its accuracy. It was also noted through visual inspection that the Bioorthogonal basis function of the Wavelet algorithm was best suited for real world images.

Abstract Format

html

Language

English

Format

Print

Accession Number

TU13646

Shelf Location

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

Physical Description

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

Keywords

Computer algorithms; Wavelets (Mathematics); Image procesings--Digital techniques; Imaging systems

This document is currently not available here.

Share

COinS