These programs are designed to register pictures of the Prokudin-Gorskii collection. http://www.loc.gov/pictures/collection/prok

• gauss : apply a gaussian filter (sigma = 0.8) to an image.

usage : gauss imgIn.png imgOut.png

• cut3 : cut a gray image into three equal height vertical slices

usage : cut3 input.tiff out1.tiff out2.tiff out3.tiff

• join3 : join three gray images of the same size to form a color image

usage : join3 in_red.tiff in_green.tiff in_blue.tiff out_colorized.tiff

• quantize : quantize an image to 8 bits

usage : quantize 0 65000 in.tiff out.png

• translation : apply an integral displacement to an image (filling-in by zero)

usage : translation 10 -20 in.png out.png

• registration : naive algorithm to register two gray images

usage : registration A.tiff B.tiff Btranslated.tiff

• pyramide : return seven images, each one zommed out by a factor two with a gaussian filter

usage : pyramide imgIn.png pyr1.png pyr2.png pyr3.png pyr4.png pyr5.png pyr6.png pyr7.png

• demo_orsa_homography : find the best homography to fit two images with the SIFT matching.

This program has been adapted from http://www.ipol.im/pub/art/2012/mmm-oh/

usage : demo_orsa_homography imgInA.png imgInB.png allMatches.txt inMatches.txt imgAWarped.png imgBWarped.png

• main : compute the four directional derivatives of an image (squared)

usage : main imgIn.png der1.png der2.png der3.png der4.png

• irani : registration algorithm based on the Irani article (Robust Multi-Sensor Image Alignment)

usage : irani imgIn1.png imgIn2.png imgWarped2.png

• homographie.sh : script which takes the original glass negative and return the registered image (with orsa-homography method)

• irani1.sh : script which takes the original glass negative, does the homography registration, and the irani warping. This may be very long for big images.