overlapping Images
ukos
I have some pictures from a microscope with a nanostage attached to it. The images are overlapping around half of the image of the successor image.
I want to have all the images combined in a single big image at their overlap and was wondering if there is a simple solution for it.
In Foto-Software the effect is often called a panorama image.
The images are all looking quite similar. Only some dirt particles are present sometimes to have a point of comparison. On the other hand I know the exact Image width from the pixel pitch size of the camera and the magnification (and tubus lenght of course). And I know the x,y position of the nanostage for each image.
My Idea was to have a simple average at the overlap and calculate to a new image. probably higher resolution of the created image is needed as the stepsize of the nanostage is probably not exaclty matching the pixel size steps etc.
As there are so many Igor-Image functions yet available I was wondering if there is something that already could do the job.
Also the problem is similar to a spectral overlap calculation which could be done via a Fermi-Dirac Statistic. Maybe fitting to this kind of weighted step function, something like A/(1+exp((X+Y)/B)) could be the answer. https://en.wikipedia.org/wiki/Fermi%E2%80%93Dirac_statistics
Has anyone done sth. similar?
1. Estimate the total output image size and allocate an appropriate output wave.
2. Start with a key image and write it into your output wave. The key image could be a corner image or even the center image.
3. Sequentially load overlapping images. In each case use ImageRegistration to register the new image to the existing edge. Once you obtain the registration transformation you can use it to write the new image into the output image with or without blending.
December 13, 2016 at 10:51 am - Permalink