An unsupervised method for quality assessment of despeckling: an evaluation on cosmo-skymed data

Goal of this paper is the development and evaluation of a fully automatic method for quality assessment of despeckled synthetic aperture radar (SAR) images. The rationale of the new approach is that any structural perturbation introduced by despeckling, e.g. a local bias of mean or the blur of a sharp edge or the suppression of a point target, may be regarded either as the introduction of a new structure or as the suppression of an existing one. Conversely, plain removal of random noise does not change structures in the image. Structures are identified as clusters in the scatterplot of original to filtered image. Ideal filtering should produce clusters all aligned along the main diagonal. In practice clusters are moved far from the diagonal. Clusters' centers are detected through the mean shift algorithm. A structural change feature is defined at each pixel from the position and population of off-diagonal cluster, according to Shannon's information theoretic concepts. Results on true SAR images (COSMO-SkyMed) will be presented. Bayesian estimators (LMMSE: liner minimum mean squared error: MAP: maximum a-posteriori probability) operating in the undecimated wavelet domain have been coupled with segment-based processing. Quality measurements of despeckled SAR images carried out by means of the proposed method highlight the benefits of segmented MAP filtering.