CS194-26: Project 1 - Images of the Russian Empire (colorizing the Prokudin-Gorskii photo collection)

For the first project of Intro to Computer Vision and Computational Photography, we had to take the digitized Prokudin-Gorskii glass plate images and generate a fully aligned color image from three color channel images.

Project Overview

The goal of this assignment is to take the digitized Prokudin-Gorskii glass plate images and, using image processing techniques, automatically produce a color image with as few visual artifacts as possible. We needed to extract the three color channel images from one image, place them on top of each other, and align them so that they form a single RGB color image. We started off with smaller jpgs and then moved onto larger, full-size glass plate images. Because the full-size glass plate images are very large, the alignment procedure needed to be relatively fast and efficient.

The Process

In the beginning, I didn't find too much trouble with aligning the smaller images. Because they were so small, it wasn't even necessary to crop them and I could just align them on top of each other. Once I moved onto the larger images, I ran into some trouble.

Cropping

Because the images are so large, I decided to crop the image to show only the center to not only avoid the poor edges of the image, but to also optimize the pixel matching

Error Calculation

Because I rescale my photo as I match photos, I had to account for some error calculations so that I'm accounting for every pixel at every level. I used SDD in order to match the pixels

Conclusion

Overall this was a fun project, that was a bit challenging. I was stuck on the pyramid portion for days before I realized I was cropping incorrectly. I ran into issues with a couple of photos (ie Emir.tiff and harvesters.tiff) becuase of their brightness, and think that edge dectection and color matching would have helped. Some things I wished I tried was automatic white balancing and automatic cropping, I started working on it but succumbed to defeat. Nonetheless, it was very fun expereince.

Results

Here are some of the images!

These are some small images.

cathedral

cathedral.jpg

Red Displacement: [1, -1]

Green Displacement: [7, -1]

monastery

monastery.jpg

Red Displacement: [-6, 0]

Green Displacement: [9, 1]

tobolsk

tobolsk.jpg

Red Displacement: [3, 2]

Green Displacement: [6, 3]

These are the larger glass images.

castle

castle.tiff

Red Displacement: [62, 30]

Green Displacement: [126, 30]

emir

emir.tiff

Red Displacement: [12, 8]

Green Displacement: [142, 46]

harvesters

harvesters.tiff

Red Displacement: [94, 20]

Green Displacement: [158, 11]

icon

icon.tiff

Red Displacement: [78, 26]

Green Displacement: [126, 34]

lady

lady.tiff

Red Displacement: [78, 7]

Green Displacement: [142, 30]

melons

melons.tiff

Red Displacement: [110, 30]

Green Displacement: [206, 46]

onion_church

onion_church.tiff

Red Displacement: [78, 33]

Green Displacement: [142, 62]

self_portrait

self_portrait.tiff

Red Displacement: [98, 4]

Green Displacement: [206, 26]

three_generations

three_generations.tiff

Red Displacement: [78, 33]

Green Displacement: [142, 46]

workshop

workshop.tiff

Red Displacement: [78, 15]

Green Displacement: [46, 30]

These are some images I got from the Library of Congress.

cliff

cliff.tiff

Red Displacement: [45, 34]

Green Displacement: [124, 66]

marsh

marsh.tiff

Red Displacement: [30, 18]

Green Displacement: [30, 22]

siren

siren.tiff

Red Displacement: [78, 9]

Green Displacement: [126, 14]

sunset

sunset.tiff

Red Displacement: [3, 1]

Green Displacement: [30, 22]

the_giver

the_giver.tiff

Red Displacement: [-4, -10]

Green Displacement: [22, -26]