“Mean-Shift” filtering to reduce speckle noise in SAR images

Speckle noise is an undesired effect in SAR (Synthetic Aperture Radar) images which degrades their quality. In this paper a filtering technique based on the algorithm “Mean Shift” is applied to reduce the speckle noise in SAR images, preserving their quality by maintaining textures, sharp edges and shapes. A image of the North-East coast of Spain is used for analyzing the influence of the parameters of the “Mean Shift” method. The quality of the filtered images is evaluated by inspection and using other objective parameters. The results are compared to those obtained using the Lee filter, which is usually considered as a standard reference for evaluating speckle filters. The proposed technique provides a greater speckle noise reduction, edges and image texture, characteristics that are very important for detecting and classifying tasks.