A statistical method for processing SAR Multichannel ATI sea surface images

A statistical analysis is presented of the process whereby weighted complex-valued Multichannel ATI (MATI) images of the sea surface are linearly combined so as to maximise or minimise the visibility of selected image features in the resulting filtered image. It is assumed that the modulated features are a result of local fluctuations in mean intensity which can be described statistically by a compound distribution. The background consists of unmodulated Gaussian scatterers. These scatterers and the modulated ones have different (and finite) Doppler spectra. It is shown that with these assumptions, maximising (or minimising) the normalised standard deviation of the filtered image maximises (or minimises) the visibility of the image modulations whatever the spectra may be or the compound distribution. This result means that it is not necessary to know the shape or pattern of the images modulations in order to enhance them. The analysis presented in this paper is aimed particularly at horizontally polarised images. A result is presented showing the effectiveness of the technique.