I'm trying to study how the texture representation is done using 2D autoregressive models. Although there are a lot of papers describing the theoretical aspects of the process, I'm finding it difficult to actually implement the algorithm to find the model parameters to represent a given texture. Is there an implementation guide or example Matlab or any other implementation code for 2D Autoregressive models for texture modelling?
You can start with this submission to FX:
https://www.mathworks.com/matlabcentral/fileexchange/5099-generate-ar1-spatial-data
And see how they implement it and what parameters they use.
From a practical point of view, I recommend the following references:
- R.M. Haralick, " Statistical and Structural Approaches to Texture ", 1979 - https://s3.amazonaws.com/academia.edu.documents/40661695/Statistical_Structural_Texture_gray.pdf?AWSAccessKeyId=AKIAIWOWYYGZ2Y53UL3A&Expires=1543315351&Signature=l2PVKo3SfcYws%2B9oQhYh0Ur1K1A%3D&response-content-disposition=inline%3B%20filename%3DStatistical_and_structural_approaches_to.pdf
- Materka et al., " Texture Analysis Methods – A Review ", 1998 - http://www.eletel.p.lodz.pl/programy/cost/pdf_1.pdf
- Hong et al., " Three-Dimensional Surface Characterization by Two-Dimensional Autoregressive Models ", 1995 - http://booksc.xyz/book/49714212/90bf0c
Thank you Eyal, Stephane for the suggestions. I will look into these. I also found, another implementation which could be useful. https://uk.mathworks.com/matlabcentral/fileexchange/29360-2d-ar-and-2d-arma-parameters-estimation
- Badges
- Science topic