Processing of Histology image with colors other than RGB

Please I need help in preprocessing this image in matlab:

The aims are to:

1. know the type of image it is? since the colors are not the usual RGB
2. remove the background of these histology images
3. separate the various color channels
4. generates counts of dead and live cells in each.

Please attached is sample of the image:

Thank you

[Star Strider]

It appears to me to be a Giemsa stained bone slice with active marrow.

Answer 1

[Image Analyst]

You can use discriminant analysis to do the color classification. Train it by outlining (say) 4 color classes

1. White
2. Dark gray
3. Light gray
4. Brown

or however many you think there should be.

Then tell it to classify all the pixels in the image to the closest class using the discriminant analysis algorithm. A full demo is attached.

[Godson Aryee]

Thank you Sir. This gives me a good starting point.

This is how it looks like;

For the 5588 pixels in class #1 ("Class1"), Red mean = 88.75, Green mean = 27.67, Blue mean = 39.68

For the 4516 pixels in class #2 ("Class2"), Red mean = 60.16, Green mean = 71.72, Blue mean = 119.04

For the 12104 pixels in class #3 ("Class3"), Red mean = 204.67, Green mean = 186.90, Blue mean = 175.69

For the 5390 pixels in class #4 ("Class4"), Red mean = 145.83, Green mean = 149.95, Blue mean = 163.94

The misclassification rate for the linear classification is 16.055 %.

The misclassification rate for the diaglinear classification is 19.134 %.

The misclassification rate for the quadratic classification is 15.833 %.

The misclassification rate for the diagquadratic classification is 19.382 %.

The misclassification rate for the mahalanobis classification is 16.046 %.