Hi features depend upon the problem what u analysis.segmentation algorithm can be applied,specifically K means algorithm.

Dear

Look the link, maybe useful.

https://afni.nimh.nih.gov/sscc/staff/rwcox/ISMRM_2006/Syllabus%202006%20-%203340/files/B_06.pdf

http://ijcsit.com/docs/Volume%205/vol5issue06/ijcsit20140506216.pdf

Article Image Segmentation and Feature Extraction

Regards, Shafagat

Feature extraction is a general term for methods of constructing combinations of the variables to get around these problems while still describing the data with sufficient accuracy. Many machine learning practitioners believe that properly optimized feature extraction is the key to effective model construction.

Different textural features are used in this work for image analysis. But, these features are estimated in a different way for the classification and segmentation applications. Since image classification is performed between images, the features are estimated from the whole image. On the other hand, image segmentation is performed within the image and hence the features are estimated for each pixel.

Features such as angular second moment, contrast, correlation, variance, entropy, inverse difference Moment, skewness and kurtosis are generally used for image classification and segmentation.

Before approaching to the type of extracted features from K means, I think we need to know what does actually kmeans mean ? We have to understand the actual purpose of K means. K means uses for unsupervised learning. So what is it ? It means we will give some data to it, and it will identify some useful relational values and group / cluster them into k groups / clusters.

Feature engineering plays a key role in K-means clustering; using meaningful features that capture the variability of the data is essential for the algorithm to find all of the naturally-occurring groups.  

I think you should watch this short tutorial for your better understanding.

https://www.youtube.com/watch?v=_aWzGGNrcic