Order of indexes when using rgb2ind

Hi there,

I have some classified images in tif format (usually 3-4 classes), which I need to import to Matlab as an indexed image. Currently I am using

img = imread(<image name>.tif);
lbl = rgb2ind(img);

Which works fine, except for one slight issue. The order of the indexes (classes) seems somewhat random. This means that I cannot compare multiple images. Even if they are all classified the same way, using the same colors for the same classes. When they are imported to Matlab, it all gets mixed up.

So my questions are:

1. Is there a (simple) way to control the order of the classes?
2. How does rgb2ind() decide the order of the classes?

Any help which could help clarify this randomization mystery is much appreciated, thanks!

Answer 1

[Image Analyst]

If you simply call rgb2ind() on each image independently, you're going to get different mappings of colors to indexes. To prevent that you have to get the colormap of one image that you trust, and apply that to all the others:

% Get the initial colormap used on image #1.
[indexedImage1, colorMap] = rgb2ind(rgbImage1);
% Now use that colormap for a different image.
indexedImage2 = rgb2ind(rgbImage2, colorMap);

In this way, all images will be colored the same way.

[Jakob Sørensen]

I will try that once I have the time, and let you know if it works. Sounds promising though... :-)

Update: Got to try it, and it works perfectly. Thank you very much, you just saved me a bunch of time!

[Image Analyst]

You're welcome. Glad I could help. Thanks for Accepting the answer.