I want to extract features like roughness,homogenity,coarseness using dwt.But I found in some answers about using glcm after dwt.Does it need?Is it impossible to extracte such features directly by using dwt.Most research paper using dwt extracted mean,standard deviation, vertical component and horizontal component features only.
Yes, you can extract the texture features directly using discrete wavelet transform. However, applying the gray-level co-occurrence matrix in advanced can enhance the quality of the extracted features (e.g. investigating features properties at different orientations and displacements). An example of this case when applied to histopathological images can be found in chapter 5 of the PhD thesis at the following URL: http://omar.alkadi.net/wp-content/uploads/Al-Kadi_PhD_Thesis.pdf
You might want to look into the Haralick texture features. These features are extracted from the co-occurrence matrix, but include energy, entropy, contrast, and homogeneity. The original paper is here: http://ieeexplore.ieee.org/document/4309314/
but these features are very common, and you can find plenty of papers and code.
Good Question i also has some doubts like this question. thanks i got an concept.
- Badges
- Science topic
- Similar topics
- Mathematics
- Algebra
- Matrix