Result Small Images:

• 00125v.jpg - River
• 00149v.jpg - Picture
• 00153v.jpg - Guy in Blue
• 00163v.jpg - Scenery
• 00270v.jpg - Fence
• 00374v.jpg - Haystacks
• 00398v.jpg - Train Station
• 00548v.jpg - Tracks
• 01167v.jpg - Building
• 01880v.jpg - Mosaic
• 01898v.jpg - Tree

• 00458u.jpg - Train
• 00500u.jpg - Suburbs
• 00911u.jpg - Two Guys and a Boat
• 01043u.jpg - Three Girls With Plates
• 01047u.jpg - Church Items
• 01657u.jpg - Woman on couch note: Not done correctly- probably because in a difference in detail in the images

• It reads a folder to generate list of images to process, namely the data/ folder
• Uses the built in edge function with 'sobel' and sensitivity of .05
• It ignores a border region since that seems to contribute negatively toward aligning images
• Will display the intermitten alignments of the images if you un-comment the imshows in the align function - quite fun to watch
• Saves all images as jpeg despite original filetype due to compatibility as well as size/ease of viewing
• Saves aligned R,G,B channels independantly as well for other use, such as easily comparing how well the images are aligned outside Matlab

• colorize_skel_2.m - Main code, vers 2 (uses recursiveness)
• align.m - Recursive Multi-Resolution code to align images
• getPad.m - calculates original edge pad/ignore size (somewhat arbitrarily calculated)
• error_sq.m - originally calculated mean squared error using raw pixel data, but after switched over to calling edge function, just calculates the difference between the edge images