I'm not sure if there is na optimum value for each feature. The value of each feature depends on the contexto of the image.

Hi once you go through the paper by haralick,,

if you find any difficulties, let me know ...we will discuss 

some important points

GLCM givea a measure of the variation in intensity  at the pixel of interest.

GLCM texture considers the relation between two pixels at a time, called the reference and the neighbour pixel

check this link too

http://support.echoview.com/WebHelp/Reference/Algorithms/Operators/GLCM_textures/About_the_GLCM_and_textures.htm

http://shodhganga.inflibnet.ac.in/bitstream/10603/9107/9/09_chapter4.pdf

hello

you can follow this  

http://www.mathworks.com/help/images/ref/graycomatrix.html

Hi Neha. With GLCM you only have a group of pattern that with help of a recognition system (like for example, a ANNs), you can use for identify or classify different characteristics of your application.

I prefer first convert RGB image to HSV. Afterwards, is necessary, to scale HSV matrix to values between 0 and 64. This do that co occurrence matrix be computationally possible. After this, you can calculate the co occurrence matrix, for H, S and V matrices. Thus, you will have three co occurrence matrix, and you can now determinate the parameters (entropy, variance.....), for each one of this matrix. It is necessary establish the correlation between parameters, for determine which of this are relevant. ImageJ (from Java) calculate this, but, if you can, is deseable do yourself the algorithms.  Neha ¿Which is your application?

Thank you all for your response.

My application for GLCM is to extract features from segmented tumor. 

Hi Neha...

First of all we have to remember that GLCM set of features are based on second order statistics... they can be used to reflect, the overall average for degree of correlation between pairs of pixels in different aspects (in terms of homogeneity, uniformity...etc). One of the main factors affects that affects the discrimination capabilities of GLCM is the separation distance between pixels... When you take the distance 1 it leads to refelct the degree of correlation between adjacent pixels (i.e., short range neighborhood connectivity). While, increasing the distance value leads to reflect the degree of correlation between distant pixels...

I think to get good picture about the inter-pixels correlation for the investigated texture, it is better to extract at least 2 sets of GLCM one for small d (e.g., 1) and other for large d (e.g., 4 or higher)...    

First question is for what ? Why you want use GLCM to analyse of these images ?

In generally, the most know way is divide your image(s) into some classes (for example one class is a background, second class contains interesting objects).

Then you can calculate coefficients (from GLCM) for both areas. Next step is a feature selection. That step can tell you what features are important, and also can remove unnecessary features. These "unnecessary features" can reduce the quality of classification, so it's important to remove. For feature selection are knows mamy methods...

After feature selection you should have only "import features" - in your case will be a set of coefficient form GLCM (for example. contrast for some direction and distance).

Anyway, this is only most basic way to solve your problem, but i think, it's will be good to start.

Thank you Loay and Jopek, for giving this much information regarding the question.

All the answers added good knowledge to me.

http://matlab.izmiran.ru/help/toolbox/images/enhanc15.html better link to understand the concept and features of GLCM

Thank Vaibhavi for your apport. Good portal. GLCM is a power tool for texture analysis. By the way, do some has work with GLCM and thermographic images? do you have had good results?

A good source: http://www.ijsrp.org/research-paper-0513/ijsrp-p1750.pdf

or simply: http://matlab.izmiran.ru/help/toolbox/images/enhanc15.html

Hello,

I have a gray scale textile images for detecting the defects. I want to segment my image using Mean Shift Segmentation Algorithm and after that want to apply the haralick features on that.

My question is that glcm is not necessary to apply with Haralick. Haralick can be applied on any segmented image.

Please suggest

The GLCM functions characterize the texture of an image by calculating how often pairs of pixel with specific values and in a specified spatial relationship occur in an image, creating a GLCM, and then extracting statistical measures from this matrix.

https://www.mathworks.com/help/images/texture-analysis-using-the-gray-level-co-occurrence-matrix-glcm.html#:~:text=The%20GLCM%20functions%20characterize%20the,statistical%20measures%20from%20this%20matrix.

I am implementing GLCM on MRI images. And also have got its features, & got the results too, but I don't understand the significance and utilization of all the features like entropy, contrast, etc. I know definition and equations, but want to know how to utilize all these features for classification and get the result.