Proj 1 - Nikhilesh Vegesna
For this project, I took reels of black and white images that were taken three times. Each time was either with a blue, red, or green filter. By combining these images, it is possible to reconstruct a color image. The process of reconstruction was made more difficult because these images were places on glass plates and were not exactly aligned. A way to fix this, once they are loaded into a digital representation, is to take each individual image and align them to each other. This was done by finding the alignments that corresponded to the smallest "distance" between images. I defined this distance metric as the frobenius norm of the difference vector between two images.
Example Images
I shifted the blue image and the red image to the green image
Cathedral.jpg: Blue Shift (-2,-5) , Red Shift (1,7)
emir.tif: Blue Shift(-24, -49), Red Shift (17,57)
harvesters.tif: Blue Shift (-16, -59), Red Shift (-3, 65)
icon.tif: Blue Shift (-17, -41), Red Shift (5, 48)
lady.tif: Blue Shift (-8, -56), Red Shift (3, 62)
monastery.jpg: Blue Shift (-2, 3), Red Shift (1, 6)
nativity.jpg: Blue shift (-1, -3), Red Shift (-1, 4)
self_portrait.tif: Blue shift (-29, -79), Red Shift (8, 98)
three_generations.tif: Blue shift (-14, -53), Red Shift (-3, 58)
train.tif: Blue shift (-6, -42), Red Shift (27, 43)
turkmen.tif: Blue shift (-21, -56), Red Shift (7, 60)
village.tif: Blue shift (-12, -65), Red Shift (10, 73)
Extra images from the collection
chapel.jpg: Blue shift (-2, -4), Red Shift (2, 6)
Suzdal.jpg: Blue shift (-1, -6), Red Shift (0, 6)