matlab images

The Sergei Mikhailovich Prokudin-Gorskii (1863-1944) was a photographer between 1909-1915 , traveling in Russia. Used a technique that involved the exposure of each scene on a glass surface using blue and red green filters. Then there was no way to print and one could see them only through special screenings . The negatives are digitized and today online ([login to view URL]). You should extract the 3 channels , put them on top of each other to form an RGB image. Auxiliary code given in the file sample_code.m. pictures given in the file [login to view URL],BGR from top to bottom . Your program should divide the image into 3 equal pieces (channels ) and align 2 in the 3rd (try different set alignment optimum result ) . For each input image should show color image and the movement vectors (x, y)which we used for alignment. The easiest way is to look efthygramisis exhaustively in a window possible movements eg [-15,15] Pixels independently in the x and y axis , find a score and keep moving with the best score . There are a number of metrics that one can use . The most basic is the L1 norm of the differences between two channels is calculated as sum (sum (abs (image1-image2))). The images that we do not have to match certainly the same prices so it is not the most appropriate way . Possible way is normalized cross-correlation (NCC), which is the inner product between two images that have been normalized to have zero mean and unit norm ( see the function of MATLAB normxcorr2). I need also comments next to each line of code to understand what you do!