Haralick function input image

Hello,

I am studying on color image processing. The data is histopatologic images and I need texture features. So that I am using Haralick functions. But I have a problem. In the image, I have neuros and they are circular shape. So I take them in the separate square black small images and then they are used in haralick function input image. is it true? or I am doing wrong thing? can The black region affect the texture results?

[Image Analyst]

I don't understand this: "I take them in the separate square black small images" What does that mean? What is a black image or region?

Also, are you talking about the Haralick texture features, like descibed here: http://murphylab.web.cmu.edu/publications/boland/boland_node26.html

[Walter Roberson]

http://www.mathworks.com/matlabcentral/answers/7924-where-can-i-upload-images-and-files-for-use-on-matlab-answers

Answer 1

[Stefan]

Hi Sibel,

if I understood you correctly: You separate the neurons from their actual image environment and st it on a black background?

Haralick feature calculation uses the Co-Occurence-Matrix (=GLCM --> http://en.wikipedia.org/wiki/Co-occurrence_matrix). This matrix represents the gray values and their adjacency.

So I am afraid you distort your results by invariant separation (different radius) of neurons an the new background (black square).