Pyvips | Best way to create large histology image of size 5000000X5000000 pixels.

Hello, @jcupitt!

I am new to pyvips, I want to create a large image. Following steps, I have to take:

1. Finding coordinates as the stage is not precise. Using NCC based matching.
2. Making a large background image to overlay each image using their global coordinate.
3. Saving it in dzi format (is it possible to view real-time stitching). I have tried all the three steps. I am stuck in last two part.

Issues:

1. When I am overlaying next image the previous image vanishes. I have tried all this method from this post . Can you also explain what is the various mode of an argument? Attached example code

import sys
import random
from PIL import Image
import numpy as np
import pyvips

output_size = 2000
bg = pyvips.Image.black(output_size, output_size, bands=3)
tile1 = pyvips.Image.new_from_file('1805_corrected.jpg', access="sequential")
tile2 = pyvips.Image.new_from_file('1806_corrected.jpg', access="sequential")
vips_image1 = tile1.embed(0, 0, output_size, output_size)   
vips_image2 = tile2.embed(1000,1000, output_size, output_size) 
vips_image=vips_image1.composite(vips_image2,["over"])
vips_image.write_to_file('vi_1.tiff')

1. My data size would be very large, what would be the best way to mosaic this images?
2. In the past, I have used Image.join function to create mosaic but the problem was when I exceed certain limit .dzsave takes lots of time and sometimes segmentation fault occurs.
3. Is there any method in pyvips to match images two find coordinates for matching?

[jcupitt]

Hello again, could you make a new issue? It’s best to have one question per issue or other people will have trouble finding answers.

Include a complete runnable program and notes on how to make input data.

[jcupitt]

Hello @shivamchaubey,

Wow, that’s a big image. If you want to paste many small images together at arbitrary positions, use insert. You will find there is a limit of about 2,000 images, set by the C stack size. You will probably have to assemble your image in sections.

If you can place the images in a grid, use arrayjoin. It can join at least 30,000 images at once.

composite is useful if your images have transparency. If you use it for images without an alpha channel, each image will just overwrite the previous one.

Yes, libvips has stuff for NCC-style matching, both 0 and 1st order. You make a series of pairwise joins along edges. It may not work for you, you’d need to give more information.