CS685 Homework 4 (Due October 14th)
J. Koˇseck´a
1. (10) Here are some examples of gaussian and derivative of the gaussian filters.
f1 = [0.1353,0.6065,1.0000,0.6065,0.1353]
f2 = [0.2707,0.6065,0,−0.6065,−0.2707]
Compute image derivatives ∂I
∂x and ∂I
∂y by convolving the image with derivative of the gaussian filters. Com-
pute and plot the gradient magnitude image and gradient orientation image (where the orientation is color-
coded by gray values). Hand in the snippet of the code which computes the two and the two images.
2. (5) Test the edge function in MATLAB and hand in images of edge maps under two different choices of
threshold and report what the thresholds are. Use the ’canny’ option to get the Canny edge detector.
3. (10) Given the output from the edge detector use the function bwlabel to compute the connected components
of the edges which are aligned with the horizontal and vertical directions. Here you need use the assumption
that the orientation of the edge at a point is ±²the horizonal and vertical directions. Hence you will run the
connected components only on pixels where the orientation satisfies this criterion. You also need to use the
result of exercise 1 for computing the gradient orientation. Hand in the snippet of the code, which computes
the connected components in the above way and the two images, one displaying the horizontal connected
components and one the vertical connected components.
There should be not for loops in this homework. Another function you will find useful is find. For problem
3 use the image provided on the web site, so you can get some meaningful answer.