Seung Chan Lim

Code

The m files haned in contains SSD, NCC and NCC with Pyramid implementations. The pyramid is automatically invoked on images that the script deems too big. You can also parameterize the amount of margin to ignore when processing the images. The pyramid can also be parameterized so that it'll stop scaling down at a certain point. There are also some extra stuff there that produces intermediate images for the fun of it (i.e. r, g, b channels separately and the non-translated composite image.

To run, simply start up matlab in the directory in which the m files reside, then type main

Example

>> main
Enter the path to the grayscale image you want processed: 'data/01656v.jpg'

Displacements

01656v.jpg
G by 0,2
R by -1,12





01167v.jpg 
G by 0,5
R by 0,12






00153v.jpg
G by 0, 6
R by 0, 13





00398v.jpg
G by 0,5
R by 0,11






00149v.jpg
G by 1, 4
R by 1, 9





00952v.jpg
G by -2, 2
R by -4, 8





00194v.jpg
G by 2,4
R by 4,8





00163v.jpg
G by 1, -2
R by 1, -4




    

These bigger images used margins between 100 and 150. Being the impatient guy that I am I also stopped the pyramid at 1/2 scale

01047u.tif
G by 20, 24
R by 32, 72





00458u.tif
G by 0, 44
R by 28, 88





01043u.tif
G by 8, -16
R by 12, 12





01657u.tif
G by 0, 56
R by -4, 116





dj.jpg
G by 8, -10
R by -44, 8

my filters



through the red filter



through the green filter



through the blue filter



processed image



whitebalance attempt



what it was supposed to look like, dammit!