In this project, we took three images that were r, g, b negatives of the original picture. These three images were greyscale and we combined them to create a colorized version of the picture.
I created two different functions, one called "align" and one called "pyramid". The "align" method was the single-scale version used to overlay the b image with the right shift of the g and r images. The "pyramid" method was the multi-scale verson to deal with larger images. It recursively worked with images from low resolution to original resolution to check for shifts to save time.
My algorithm did not work on the emir.tif image as well because it was not able to find the right shift for the image as it was too bright.
gb offset: (-9, 2)
rb offset: (-11, 3)
gb offset: (-23, 12)
rb offset: (-41, 23)
gb offset: (-17, 7)
rb offset: (-31, 6)
gb offset: (-27, 8)
rb offset: (-50, 11)f
gb offset: (-19, 4)
rb offset: (-42, 5)
gb offset: (-13, 2)
rb offset: (-14, 2)
gb offset: (-7, 1)
rb offset: (-12, 0)
gb offset: (-9, 14)
rb offset: (-9, 18)
gb offset: (-3, 0)
rb offset: (-6, -1)
gb offset: (-19, 6)
rb offset: (-38, 4)
gb offset: (-26, 2)
rb offset: (-52, 15)
gb offset: (-19, 9)
rb offset: (-38, 13)
gb offset: (-38, 1)
rb offset: (-27, 11)