Revisiting the effect of spatial resolution on information content based on classification results

Abstract

Polarimetric Synthetic Aperture Radar (PolSAR) images are an important source of information. Speckle noise gives SAR images a granular appearance that makes interpretation and analysis hard tasks. A major issue is the assessment of information content in these kinds of images, and how it is affected by usual processing techniques. Previous works have resulted in various approaches for quantifying image information content. In this paper, we study this problem from the classification accuracy viewpoint, focusing on the filtering and the classification stages. Thus, through classified images, we verify how changing the properties of the input data affects their quality. The input is an actual PolSAR image, the control parameters are (i) the filter (Local Mean, LM, or Model-Based PolSAR, MBPolSAR) and the size of their support, and (ii) the classification method (Maximum Likelihood, ML, or Support Vector Machine, SVM), and the output is the precision of the classification algorithm applied to the filtered data. To expand the conclusions, this study deals not only with Classification Accuracy but also with Kappa and Overall Accuracy as measures of map precision. Experiments were conducted on two airborne PolSAR images. Differently from what was observed in previous works, almost all quality measures are good and increase with degradation, i.e. the filtering algorithms that we used always improve the classification results at least up to supports of size 7 × 7.