Difference key is not going to work, sorry for the bar idea. Try placing the slabs over a flat surface, a board or something opaque and that has a flat color (black, white, green or blue screen) so that you have a clear separation between what is marble, and what isn't. if you manage to automate finding and masking the edges of the slabs, then finding control points and stitching the pictures should be easier to automate. On Sunday, August 4, 2024 at 1:23:56 AM UTC-7 Claudio Rocha wrote:
> > I think that you can do this in hugin through scripting. > > The basic workflow would be like the way scanned images are stitched > together: > > Read: > https://hugin.sourceforge.io/tutorials/scans/en.shtml > > The idea is to treat each image as a different lens in the hugin file, and > stitch them as mosaic plane using a rectangular projection. > Think of the camera as if it was moving on a rail over the stationary > slabs. For the optimization, hugin should deal only with movement in X and > Y axis (and maybe rotation). > > For the camera original photos I would suggest to *use a format that is > not jpeg*, shoot raw and work with images with a bit depth of 16 bits per > color. More on that below. > > The first image you need is a photo of the empty conveyor belt without any > slabs. Just the empty space using the same lens and light. > > Use that first image to automatically subtract the background from all of > the images with slabs of marble. > Generate what is called a "difference mask" or a "difference matte", and > use that to add black, white or alpha around the slabs of marble. > > For automated scripting that can be done with imagemagik > > https://imagemagick.org/Usage/compose/#changemask > > This will work best if the original files have no jpg compression > artifacts. Shoot raw and convert to 16 bit tiff or png files. > Having images without a background will make it easier for Hugin to create > control points that are only within surface of the slabs, and not in the > conveyor belt. > > Once you have a series of images that contain only the slabs, then you can > run a script. > Here's an example in bash, but I'm sure porting it to python should not be > difficult. > The goal of the script is to process a series of images on the current > directory and stitch them together. > > > > > > > > > > *#! /bin/sh# hugin command tools script to stitch images as a mosaic > plane# use of fov >= 10 should be OK, could simply set FOV=10# set fov > valueFOV=10#create a .pto project using all jpg files in current > directory**pto_gen > --projection=0 --fov=$FOV -o project.pto *.JPG* > #modify accordingly for tiff, or other formats > > *# for loop that counts how many images exist, and then assigns a new lens > to each one.* > > > > > > > > > > > > > > > > *#modify for different image formats if needed.for I in $(seq 1 $(ls -f > *JPG | wc -l)) do pto_lensstack -o 01_project.pto --new-lens i$I > project.ptodone#find control pointscpfind --prealigned -o > 02_control_points.pto 01_project.pto#clean control pointscpclean -v > --max-distance=1 --pairwise-checking -o 03_clean_cp.pto > 02_control_points.pto# Optomization is set to rotation, X and Y > coordinates.* > *pto_var -o 04_set_optim.pto --opt r,TrX,TrY,!r0,!TrX0,!TrY0 > 03_clean_cp.pto* > *# if rotation creates problems then change to *-opt *TrX,TrY,!TrX0,!TrY0* > > > > > > > *#calculate the panorama using all the control points and set > parametersautooptimiser -n -o 05_auto_optim.pto 04_set_optim.pto#set the > projection to rectilinear and calculate croppingpano_modify --projection=0 > --fov=AUTO --center --canvas=100% --crop=AUTO -o > 06_final_stitch_project.pto 05_auto_optim.pto* > > You can use the hugin user interface to check each of the projects along > the way, and understand the logic behind it, and troubleshoot any problems. > > On Saturday, August 3, 2024 at 4:43:33 AM UTC-7 [email protected] wrote: > >> Hello, >> >> We are an Italian company named : Iride www.iridefoto.it >> >> we have developed an automatic photography system for marble slabs. >> We are taking photographs of marble slabs passing through the polishing >> line. >> >> The photography position is fixed, meaning we always have the same shot. >> >> We have to compose the final marble slab starting, for example, from >> about ten photographs. >> We take a picture every 60cm and the final slab will be for example >> around 350cm. >> The pictures have overlapping areas to facilitate the joining. >> >> We have tryed to crop the photographs selecting only the marble part >> before to perform the final algo. >> Basically we need to have a photo stitching solution. >> >> We will prefer to have a solution with Python as our script is already >> written with this language. >> >> The script doesn't have to a "real-time one" >> We can process the pics and after "minutes/hour" produce the result for >> every slab >> >> We have a demo of the images here : >> >> https://drive.google.com/drive/folders/1pkYErc_4Kdzajqlno6zACtNJRZJJ4GWp?usp=sharing >> >> We have played with this project on Git: with no good result. >> https://github.com/OpenStitching/stitching >> >> >> This is a job request that provides compensation to anyone who can >> provide a software solution. >> >> Many thanks for your help. >> www.iridefoto.it >> > -- A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ --- You received this message because you are subscribed to the Google Groups "hugin and other free panoramic software" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/hugin-ptx/360c76dd-d1f4-4c01-90c7-5799ccaf12f9n%40googlegroups.com.
