What kind of significant feature for these image

Hi all I want to take your suggestions what kind of significant or discriminative features for this image can be use like gray level or binary level. I just try some traditional ones like GLCM but not effecive so your suggestions. There is four different image there

[Walter Roberson]

Please attach the image.

[Ahmet Cecen]

What is the purpose of this feature extraction?

[ali]

classification like aging test or compare the result with experimental applications

[Ahmet Cecen]

Do you care if you can put a name to these feature and make sense of them (say average fiber length), or are you ok with features that you cannot relate to (no physical meaning), but that would still work for classification?

[ali]

I have one sample for each one. So first I think, each image should be divide into non overlap sub images. The feature name for me important for example regularity, legnth, irregularity or homogeneity whatever think. The well known algorithm is not work on them like GLCM(not useful results) or LBP(local binary pattern) not so effecive. So i want to combine some features maybe found new property i dont know still looking.

[Ahmet Cecen]

Only an expert in your field can help you with physically meaningful features, you likely won't find any help here.

[ali]

Anyway, some good advises comes through there whenever I share topic there. Because all surfaces have meaningful description. And also expert for these images only know its experimental data. So image analyst has better view for this.