The code attached is all about skin tone segmentation in 3 integrated color spaces, the problem is the code produces no segmented image as o/p and I can not find the problem yet. Anyhelp please

The code attached is all about skin tone segmentation in 3 integrated color spaces, the problem is the code produces no segmented image as o/p and I can not find the problem yet. Anyhelp please

[Ahmed El Sheikh]

Thresholds are used in a paper and its positive detection rates are 90+% And even when i used r>50 as my threshold it also showed nothing. Is my approach correct syntax & logical wise?

Answer 1

[Image Analyst]

You chose bad thresholds and you need to adjust them. Try the Color Thresholder on the Apps tab of the tool ribbon.

[Ahmed El Sheikh]

Even when i tried other thresholds nothing was shown as o/p I want to know what is the problem and it is not about thresholds i guess My syntax/logic is correct?

[Image Analyst]

You can try one of mine, attached. I've just tried it on the demo image so I don't know how well it will work for your images.

Also attached is an m-file that plots skin colors as measured from a clinical study so you can see the banana-shaped gamut and realize why no box segmentation method (like specifying low and high thresholds for each color channel) will work perfectly.

[Ahmed El Sheikh]

The pt is i am segmenting on 3 color spaces not a single color space

[Image Analyst]

The image might be floating point. Try this:

imshow(uint8(rgb));

or

image8 = uint8(255 * mat2gray(rgb));
imshow(image8);

[Image Analyst]

Ahmed, did that work?