Quantitative measurements on despeckling of SAR clutter amplitude : An experiment on MSTAR data

Synthetic aperture radar (SAR) images are inherently affected by a multiplicative noise known as speckle that makes visual and automatic interpretation a difficult task. Till date, many filters have been developed for speckle reduction in SAR images. In this paper, a comprehensive study on various despeckling filters such as Lee, Kuan and Enhanced Lee is presented. In addition, the maximum a posteriori (MAP) estimate of texture amplitude for Γ-MAP filter is derived. Furthermore, quantitative evaluation of these filters is made on moving and stationary target acquisition and recognition (MSTAR) clutter amplitude data using classical measures such as equivalent number of looks (ENL), speckle suppression index (SSI) and few recently proposed statistical measures i.e. mean preservation index (MPI) and mean preservation speckle suppression index (MPSSI). Experimental results illustrate the effectiveness of these filters in terms of both noise reduction and detail preservation at the same time.