E186 Project 2 (due March 6)

This project contains three parts, image sharpening, more on edge detection, and color image enhancement. Along with the images you are required to submit (indicated in each of the following problems), also submit a text file of a brief description of what you did for each of the problems. Since image enhancement is very subjective and application oriented, you need to be creative while designing the algorithms for the best effects.

As always, add your last name in front of the image names specified below and submit them to ``/students/proj3''.

1.

More Edge Detection

Use the following methods to detect edges in /e186/imagedata/panda.pgm. The purpose here it to get the a binary line drawing of the panda in the image.

(a)

The Laplace operator combined with zero-crossing.

(b)

LoG and zero-crossing. Try different $\sigma$ in the Gaussian filter to achieve best result.

(c)

DoG and zero-crossing. Try different combinations of $\sigma_1$ and $\sigma_2$ in the two Gaussian filters to achieve best result.

(d)

Canny method (available in /home/adder/classes/e186/programs/canny). Try different parameters ($\sigma$ and the two thresholds) and their combinations to achieve best result.

Submit these results as images 1, 2, 3 and 4.

2.

Still More on Edge Detection

As I showed you in class, it is effortless for our visual system to detect the edges in the image /home/adder/classes/e186/imagedata/noisy_box.pgm. But if you look at the cross section of the image (/home/adder/classes/e186/imagedata/noisy_box_profile.pgm), you will see it may not be so effortless for a computer program to detect the edges. Now do whatever it may take to detect the edges (a square) so that they are represented by lines with single-pixel width. This is obviously a multi-step process. Submit the result as image 5. Could you postulate how our visual system may be doing this?

3.

Image Sharpening

Design several different high-pass masks (including high-boost) for image sharpening. Here the purpose is to enhance the edges and other details such as textures, but we still want to preserve the basic contents of the image. Try kernels of different sizes for best result. Apply these masks to /e186/imagedata/blurred_cat.pgm and an image of your choice and submit the best result for each image. name them image 6 and 7.

4.

Color Image Processing

The purpose here is to improve the appearance of a color image of your choice. There are quite a few color images in ``/e186/imagedata/'' (all those images with .ppm extension). You should choose one of those images with poor contrast, poor color saturation (such as "NorthSea.ppm") and distorted colors so that the effect of your processing will be most significant. Submit the color image of your choice as image 8.

Two pairs (four) functions are provided for RGB-HSI conversion in /e186/programs/rgbhsi.c. One pair is based on the RGB cube color model and the other is based on the hexcone HSV model. Go through the code to see how the conversions are implemented before you use it. In particular, you should notice that the intensity in the function is represented by a value between 0 and 1 (instead of 0 and 255!). Try both methods for the following and submit the better one for each. Comment on the difference (or similarity) between the two methods, if any.

(a)

Increase the dynamic range of the intensity I of the image by linearly stretching its intensity histogram while keeping the hue and saturation unchanged. Submit the result as image 9.

(b)

Increase the saturation S of the image while keeping their hue and intensity unchanged. Submit the result as image 10.

(c)

Modify the hue H by changing the hue angle. Submit the image with increased hue (e.g. by 45 degrees) as image 11 and the one with decreased hue as image 12.

(d)

Modify all three parameters H, S, and I of the image to make it most beautiful (good saturation, reasonable contrast and proper hue)! Submit the result as image 13.