ECE 537 Digital Image Processing
Homework Assignment #8, Image Segmentation
Prof. Hintz
April 13, 2008
Due: April 16, 2008
Select two gray scale images of your choice as follows:
• one with predominantly straight edges
• one with predominantly curved edges
For each of the images
1. Choose one of the three standard (Sobel, Prewitt, and Roberts) edge detectors, apply it,
and print the results. Use these two images as baselines for the following comparisons
2. Apply the LoG edge detector to the images and print. Compute and display the RMS
difference between your standard edge detected image from 1 above.
3. Apply the Canny edge detector to the images and print. Compute and display the RMS
difference between your standard edge detected image from 1 above.
4. Apply a local threshold to the original images (noise riding threshold) and display the
resulting image.
5. Apply a global threshold to the original images by:
a. Assume the gray scale distribution of the image is Gaussian.
b. Compute the mean and variance of the image.
c. Threshold the image at 1 standard difference above and below the mean (set
values below the low threshold to the low value; set values above the high
threshold to the high value), print the image.
d. Scale and histogram equalize the resulting thresholded image to fill the entire
amplitude dynamic range of the original image.
