ECE 563 Assignment 9 (due 10 March 2004)

Individual Work

1. Use the options of the Matlab edge function to prepare an array of images including the original image and five different "edge" versions.

2. Compare edge images generated by "minmax" filters to those produced by laplacian filters and those produced by the grayscale morphological gradient described on p 366-369 of the textbook (see figure 9.23(d).

Group Projects

1. Find the best-fitting gaussian for binomial filters of order n=2..10. Show plots of the (continuous) gaussian and the (discrete) binomial filter weights for each order.

2. Demonstrate the technique of unsharp masking on images of your choice.

3. Write a Matlab routine to generate an image containing the standard deviation at each pixel of the base image, e.g. out = imdeviation(in,n) where in is the input image and n is the size of the filter (try n= 5 .. 11. Test your function of the fruit images and on other images of your choice. Observe whether this operation offers any further insight into the problem of segmenting objects in an image. Note you should be able to do this without using for loops, and that you can use imfilter (or similar routines) in the process.

4. Investigate whether edge-finding offers a possible method of identifies individual pieces of fruit from the images of fruit studies previously.

5. Prepare Powerpoint tutorials on the following topics (see Using Deblurring in IPT Help):
   1. Group 1: blind deconvolution, deconvblind
   2. Group 2: Lucy-Richardson, deconvlucy
   3. Group 3: Regularized filter, deconvreg
   4. Group 4: Wiener filter wiener2 and deconvwnr.
   Provide as many details as you can find (examine the matlab source) about each method, and give examples of its use.

Maintained by John Loomis, last updated 2 March 2004