Texture-Based Analysis Techniques for the Classification of Radar Images

Useful information for radar image interpretation and classification may be derived from a texture-based analysis, since their intrinsic spatial variability or texture is a valuable feature for discrimination among different regions. In the paper four different texture classification algorithms are described. The first two follow a structural approach to texture analysis and are based on the spectral texture properties detected from the Fourier spectrum. The last two are based on statistical texture properties, regarding texture as a two-dimensional random field defined by a suitable autoregressive model. These algorithms have been applied to meteorological images from a dual linear-polarisation radar and provided good results.