How to effectively classify images based on color feature?

I have set of currency images which i am classifying based on their RGB value. Due to some similarity in pixels of different class images I am not able to classify them properly..

Please suggest some ways to improve Thanks

Answer 1

[Image Analyst]

Just take the mean LAB value. Convert RGB into LAB with applycform. Get the right transform and be able to handle illumination differences by using an x-rite Color Checker Chart so that you can calibrate your intensities and colors. The notes almost certainly have different LAB means. If they don't then look at the means in smaller areas, say, divide the note up into quadrants and look at the means there. It should not be too hard to figure out which image is which currency note.

[swati kamat]

i want the code for this can u help m

[Image Analyst]

From the help for makecform:

Convert RGB image to L*a*b*, assuming input image is sRGB.

rgb = imread('peppers.png');
cform = makecform('srgb2lab');
lab = applycform(rgb,cform);