Информативность систем текстурных признаков для классификации спутниковых изображений с высоким пространственным разрешением

Quality of 4 popular texture features extraction methods and their application in the field of classification of hi-resolution satellite images are studied. The discussed methods are: Gray Level Difference (GLD), Gray Level Co-occurrence Matrices (GLCM), Gaussian Markov Random Field (GLRM) model and Simultaneous Autoregressive (SAR) model. Quality of combined sets of textural features and minimization of these sets are also studied.