|Title||Gaussian MRF rotation-invariant features for image classification|
|Publication Type||Journal Article|
|Year of Publication||2004|
|Authors||Deng, H., and D. A. Clausi|
|Journal||IEEE Transactions on Pattern Analysis and Machine Intelligence|
|Pagination||951 - 955|
|Keywords||Algorithms, anisotropic circular Gaussian MRF model, Artificial Intelligence, Automated, Computer Simulation, Computer-Assisted, discrete Fourier transform, discrete Fourier transforms, Gaussian MRF rotation-invariant features, gray level cooccurrence probability features, image classification, image enhancement, image interpretation, image texture, Laplacian pyramid, Markov Chains, Markov processes, Markov Random Field models, Models, Normal Distribution, Pattern Recognition, Rotation, rotation-invariant texture features, Statistical, statistical improvement, texture rotation|
Features based on Markov random field (MRF) models are sensitive to texture rotation. This paper develops an anisotropic circular Gaussian MRF (ACGMRF) model for retrieving rotation-invariant texture features. To overcome the singularity problem of the least squares estimate method, an approximate least squares estimate method is designed and implemented. Rotation-invariant features are obtained from the ACGMRF model parameters using the discrete Fourier transform. The ACGMRF model is demonstrated to be a statistical improvement over three published methods. The three methods include a Laplacian pyramid, an isotropic circular GMRF (ICGMRF), and gray level cooccurrence probability features.