[Resources] [Terminology] [Units] [MCQs]


   

Edge Detection


30 min.
Objectives:
At the end of this practical you should understand the role of edge enhancement and thresholds in edge detection. You should also gain an appreciation for the difficulty of evaluating feature detection algorithms and the potential use of ROC curves.

Terminology:


See:
Connectivity
Contrast
Convolution Enhancement
Receiver Operator Characteristic
Read the introductory sections in the MATLAB user guide regarding file I/O and displays. Familiarise yourself with the MATLAB "HELP" facility.

Run the matlab "demos" and select the "edge detection" demo from the "Image Processing" toolbox.

Try out the various edge detectors on the images provided and try to asses their relative performance on the basis of noise, connectivity and completeness of identified edges. Is there a method which performs best on all or most of the images?

Now try manually adjusting the edge threshold. Is it possible to improve upon the initial detection results with a manually selected threshold?
How does the threshold relate to the "contrast" of a detected edge?
How would it be possible to determine whether performance was due to a change in the filters used or the threshold?

Useful resources

MATLAB home page
Image and Video processing solutions page
Image Processing Toolbox product documentation


(c) Imaging Science and Biomedical Engineering 2000 [paul.bromiley@man.ac.uk]

Valid HTML 4.01!