More Research Statistical Properties Under The Influence Of Sas Image

Draw lessons from successful application of Synthetic Aperture Radar (SAR) in ground imaging, synthetic aperture technique was introduced into sonar field in1960s, resulting a new high resolution underwater acoustic imaging sonar--Synthetic Aperture Sonar(SAS). It synthesizes an aperture by storing successive echoes obtained from a moving platform and by processing the results as if they had been obtained from a multi-element array enables a high azimuth resolution to be obtained from a physically small array. High-resolution SAS image is a coherent record of scattered acoustical energy and often resembles a photograph of the seafloor, with recognizable structures such as sand ripples, beds of seagrass, rocks or man-made objects. High-resolution SAS image statistics is important for a number of practical applications, especially classification of various types of seafloor, mine detection and the laying of gas or oil pipelines.Previously SAS image statistics is modeled by the Rayleigh distribution based on the central limit theorem(CLT). However, both heterogeneity and spatial correlations violate the independent and identically distributed condition of the CLT, while the small sonar resolution scales achieved by SAS also violate the infinite-number-of-scatterers condition. Thus, high-resolution SAS image statistics is no longer well fit by the Rayleigh distribution. On this basis, the K-distribution is proposed to describe the statistics of SAS images in this paper. This study will help to reduce multipath contamination and will be able to improve the SAS image quality greatly.1. In this paper, the K-distribution is shown to represent statistics of SAS images without multipath contamination. High-resolution SAS imagery often exhibited significantly non-Rayleigh, heavy-tailed statistics, characterized by the K-distribution with a low shape parameter. And the shape parameter is proportional to the number of scatters or patches within a resolution cell, proportional to the beam width of the receive array and inversely proportional to the transmit waveform bandwidth. Use of the K-distribution is justified from the results comparison of the K-distribution and Rayleigh distribution, utilizing the simulation SAS images. 2. statistics of contaminated SAS image is modeled by a K-K mixture model, including two K-distributions related to the direct-path and the multipath return separately. A K-distribution is proposed to approximately describe multipath propagation. Justification for this approximation was found in both the matching of higher order moments and in a limiting distribution. The impact of multipath on SAS image statistics is explained by using the K-distribution shape parameter as a metric. Both theory analysis and simulation data prove that multipath contamination produces more Rayleigh-like statistics than direct-path-only return, decreasing the pdf tails and increasing the shape parameter.3. This paper proposes a modified estimation method of the parameters of the K-K mixture model:EM-MOM (Expectation Maximization-Method of Moment) Algorithm. Application of simulation data verifies the effectiveness of the method.