Statistical scattering model for high-resolution sonar images: characterisation and parameter estimation

The statistical mechanism ruling the reverberation from the seabed in high-resolution sonar devices is investigated. The signal received from a sonar cell is assumed to arise from the coherent sum of a random number of independent elementary contributions. Under proper conditions regarding the surface roughness and the incidence angle, the observables can be modelled as compound-Gaussian random variables with a nonuniform distribution of phase. A general expression has been found for evaluating the first- and second-order probability density function of the complex observables, and the case corresponding to a gamma distributed texture is investigated. The authors also propose a technique for estimating parameters of the generalised K distribution based on approximating the design distribution with a neighbouring family. Results have demonstrated that the algorithm achieves good performance within the examined range of parameter values