Information such as contrast, entropy, dissimilarity index and variance are computed to know the properties of any image, but how these values are used to for further classification or feature extraction?
Image Texture Classification using Gray Level Co-Occurrence Matrix Based Statistical Features
In this paper, a novel texture classification system based on Gray Level Co-occurrence Matrix (GLCM) is presented. The texture classification is achieved by extracting the spatial relationship of pixel in the GLCM. In the proposed method, GLCM is calculated from the original texture image and the differences calculated along the first non singleton dimension of the input texture image.
Article Image Texture Classification using Gray Level Co-Occurrence ...
For starters Harshit , I would suggest reading William Pratt ,
There it is defined as texture (with relevance to ) coarse or fine details (based on neighborhood of pixels repetition) thus allowing the classification on the size of the (spatial) neighborhood. e.g: Wool and cotton (the fabric in terms of strands)
Using GLCM values a secondary calculation that defines the variance or SD etc the spread indicator(s) is used to differentiate texture (google : Haralick's spread indicator )
Dear researcher
You can use the following process for feature extraction based on GLCM:
1- Compute GLCM for input image (different size,angles,etc)
2- Computer value of entropy, energy, skew, mean, etc from provided Gray-level co-occurrence matrix
3- Consider a numeric feature vector with N dimensions. Each dimension shows the value of one of the computed measures
For example final extracted feature vector is :
(For example)
4 - Finally you can classify textures based on these feature vectors and using a suitable classifier such as KNN, KD, SVM, DT, etc.
Thanks Shervan Fekri.
I want to know methods for glcm computatoion and optimize glcm vector.
- Badges
- Science topic
- Similar topics
- Computer Science
- Images