Use this URL to cite or link to this record in EThOS:
Title: Edge detection of textured images using multiple scales and statistics
Author: Williams, Ian Anthony
ISNI:       0000 0001 3569 0804
Awarding Body: Manchester Metropolitan University
Current Institution: Manchester Metropolitan University
Date of Award: 2004
Availability of Full Text:
Access from EThOS:
Texture is often the discriminator for different regions of an image. It can allow a region, or an object's edges to be represented as a difference in the pixel texture properties, as opposed to a difference in intensity. When analysing images with significant levels of noise, clutter or texture, the inadequacies of many common edge detectors has been noted. Where these traditional techniques fail, texture based edge detection proves more appropriate. In this work novel statistical edge detectors particularly suited for textured images are designed, presented and analysed. These are based on two-sample statistical tests which are used to evaluate any local image texture differences and by applying a pixel region mask to the image analyse the statistical properties of that region. The technique is enhanced further by combining multiple sized masks and multiple-statistical tests using a neural network traineq to classify many edge types using outputs from this technique. This results in a more robust and consistent detection of texture edge profiles. An analysis of these novel techniques shows an improved performance over the current standards in edge detection, namely, the benchmark Canny filter. This work further investigates the inadequacies of current edge detector evaluation metrics, and as a contribution to this field presents a novel grey-scale comparison metric for objectively evaluating edge detection performance.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available