Project 0

This project was to learn how to overlay images of different color scales. We took real world data of separate color scale plates to make a

full colorized image. My approach was to first crop out the white borders for easier aligning. I had two methods for aligning: the naive exhaustive

search and the gaussian pyramid search method. The first method just takes a window of displacements and tries every one of them to see which displacement

works the best. All of these images were aligning the green and red plates to the blue plate. The gaussian pyramid way first does a gaussian blur over

the image and then rescales the image by half. This continues until the image is under a certain size. Then it begins trying to find the best displacement.

The best displacement is propogated down the pyramid and is used as an estame for the next best displacement. For the photo of emir, I had to align the

blue and red plates to the green plate. The metric for alignment I used was zero-normalized cross correlation.

Green disp: [5, 2] Red disp: [12, 3]

Blue disp: [-49, -24] Red disp: [57, 17]

Green disp: [59, 16] Red disp: [124, 13]

Green disp: [40, 17] Red disp: [89, 23]

Green disp: [48, 9] Red disp: [112, 11]

Green disp: [82, 10] Red disp: [129, 13]

Green disp: [-3, 2] Red disp: [3, 2]

Green disp: [51, 26] Red disp: [108, 36]

Green disp: [78, 29] Red disp: [176, 37]

Green disp: [53, 14] Red disp: [112, 11]

Green disp: [3, 2] Red disp: [6, 3]

Green disp: [42, 5] Red disp: [87, 32]

Green disp: [65, 12] Red disp: [137, 22]

Green disp: [52, 0] Red disp: [104, -12] Green disp: [22, 21] Red disp: [60, 29] Green disp: [71, 39] Red disp: [148, 63]