Validation of Texture Measures for Cumulus Cloud Image Classification

M. Maskey and T.S. Newman (USA)


Texture, MRF, Gabor filters, GLCM, content-based image retrieval, pattern classification


A texture measure validation study is described. The goal of the study is to determine objectively which of three classes of texture measures are most suitable for classifying images of cumulus clouds. The paper compares the cluster quality of texture feature vectors that are extracted using Markov Random Fields, Gray Level Co-occurrence Matri ces, and Gabor filters. A large set of real images is used for the validation. In addition, the findings are applied to evaluate two artificial cloud synthesis techniques.

Important Links:

Go Back