作者: Gerardo Di Martino , Giovanni Pecoraro , Giovanni Poggi , Daniele Riccio , Luisa Verdoliva
DOI: 10.1007/978-3-642-41184-7_2
关键词:
摘要: Image databases and benchmarks are precious tools to assess the quality of competing algorithms fine tune their parameters. In some cases, however, cannot be captured by a single measure, several them, providing typically contrasting indications, must computed analyzed. This is certainly case for SAR despeckling field, also because lack clean reference images, which forces one compute measures interest on simple canonical scenes. We present here first results an ongoing work aimed at selecting suitable combination benchmark techniques rank them. The full validation proposed methodology will require involvement reasonable number expert photo-interpreters large-scale experimental campaign. Here, we only sample experiment provide insight about approach.