Forays into color correction

I posted a while back about using RGB LEDs and a monochrome camera to acquire color images, and while this works pretty well, one thing that has bothered me about it is getting the right color balance of the images. That is, does the color image captured by this process correspond to the same image you would see under white light illumination?

Datacolor has released a calibration slide and software tool for doing this color calibration, but I want to see if we can do it cheaply and build a tool for color correction into Micro-manager. To that end, I got a 35 mm slide designed for calibrating flatbedĀ scanners and imaged it on our microscope, with RGB LED illumination, and white balanced by dividing by flatfield images taken with no sample present..

Coloraid.de calibration slide imaged with RGB LED illumination and stitched together from 77 images with a 4x / 0.2 objective. Click to open the full size image at Gigapan.

Coloraid.de calibration slide imaged with RGB LED illumination and stitched together from 77 images with a 4x / 0.2 objective. Click to see full resolution image at Gigapan.org.

In other words, we’ve turned our microscope into the worlds most expensive 35 mm slide scanner, and in the process acquired a 280 megapixel image of that slide. It’s sort of interesting, actually – you can very clearly see the grain in the film. I’ve written an ImageJ macro to extract the RGB values of each square, and now I need to figure out how to transform them into some standard color space. Unfortunately, color spaces are complicated (to me anyway); for example, see the Wikipedia pages on XYZ or Lab color spaces. But I hope to make some progress on this soon….

4 thoughts on “Forays into color correction

    • Thanks for the suggestion – ArgyllCMS worked great. I’ll put together a post describing the process soon.

  1. hi kurt,

    can we use rose bengal or other related dyes of different concentrations for detecting PMT linear behavior? if not, please tell any other protocol for the same.

    thank you

    • I don’t have a lot of experience with measuring the linearity of PMTs but I don’t see why you couldn’t use a sample like this. Lots of fluorescent samples will probably work for measuring PMT linearity, I would guess.

      Also, this is a weird post to attach this comment to.

Leave a Reply

Your email address will not be published. Required fields are marked *