codingprolab@gmail.com

$30.00

Order Now
Category: EECS 442

Description

5/5 - (4 votes)

1 [55 points] Harris Corner Detection

In this problem, you will be using the images I1.png and I3.png.

(a) [20 points] Implement the Harris corner detector by computing the second moment matrix M at each

pixel location. Choose an appropriate value for the size of the window W. Refer to the class notes for

the notation. Return the source code of your implementation. Clearly indicate the key steps of the

algorithm as well as the choice of parameters in the source code.

(b) [20 points] Compute the eigenvalues of M at every pixel locations. Compute the ellipses associated to

such eigenvalues as discussed in the class notes or in FP page 184. Superimpose such ellipses on the two

images similarly to Figure 8.17 or 8.18 in FP. Repeat this step by using different sizes of the window W.

Return zoomed details of images I1 and I3 with superimposed ellipses for different values of the window

size. Comments your results.

(c) [15 points] Use the corner response R to assess whether a corner has been detected or not. Superimpose

corners with R > T on the two images. Explore a few different values of T. Return the images I1 and

I3 with superimposed corners for different values of T. Comments your results.

2 [45 points] Blob Detection

(a) [30 points] [Blob Detector ] Implement a blob detector based on the Laplacian operator and characteristic

scale concept introduced by Lindeberg.

(1) Apply your blob detector to the image 3.bmp.

(2) Plot each blob as a circle with radius given by the scale-space analysis (e.g., see figure above). Use

a Threshold T to suppress blob responses. Report your choice of T and comment on its value.

Note: Use MATLAB function show all circles.m to plot the circles.

(b) [15 points] [Description] Compute the radius of each blob (in pixels) and report the histogram (distribution) of radii computed for the whole image. Show numerical values. Normalize radii from 0 to 1

before plotting histogram. Use 10 bins in your histogram. Use Threshold T = 1500.

1

WhatsApp us