A comparison of clutter texture properties in optical and SAR images

When natural clutter textures are imaged with either incoherent optical or coherent radar sensors theory suggests that the resultant image textures should contain a common contribution corresponding to the incoherent imaging of underlying cross-section fluctuations. In this paper we examine the theoretical relationship between such images in terms of their single-point moments and the intensity autocorrelation function. We demonstrate that a correlated noise model based on gamma-distributed noise is consistent with both image textures. Based on this finding we use the incoherent optical image to simulate a set of SARlike textures which are statistically indistinguishable from the original SAR texture. This represents a significant step in the ability to combine information from different sensors.