I have attempted doing modal analyses by image analysis but have not felt confident enough to use any of them. Part of the difficulty is that you need a very large image to get a representative area. Also, transmitted light images are bad due to uneven illumination, color differences, variations in interference colors etc. A possible solution is to use SEM images in backscattered electron mode. Each mineral should have a distinct grayscale level. Except for zoned crystals. And grain boundaries. This also requires a polished thin section and time on an SEM. Another method, even more costly is to use an automated probe to count thousands of points using WDS. It takes hours but you can do it and it runs unattended. However, you can only analyze for about 5 elements (+ backscattered levels) depending on how many spectrometers you have. Sometimes that is not enough to distinguish all the phases.
A traditional 1000 point count on a standard thin section takes about 20 min if you know what you are looking at. Its not fun but it works. I would save the image analysis for a time when you need to do very fine work such as getting changing mineral modes across a vein. Then it might be worth the time.
John Tacinelli
Earth Science Instructor
Rochester Community and Technical College
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|