Resolution Sar Clutter Analysis And Application Of Statistical Characteristics

Clutter is the inherent environment for radar signal process. The research on statistical character of clutter is very important. It's helpful for understanding the scatter character of clutter, directing designing and improving algorithmic of object detection. It's also effective in directing SAR image classification and segmentation. As the resolution increasing, the statistical character of clutter is more and more complex. The paper researches deeply on statistical analysis of high resolution SAR clutter.At first, some inherent characters of SAR image are introduced, including speckle, shadow, perspective and shrink. The effect of system parameters and surface characters is also introduced. And the representations of typical clutter in image are also provided. The most important analysis is on speckle mechanism, model and the product model of SAR echo.Then, some experiential models which are based on experiments are provided, and the applicability is discussed by simulation. The research indicates that these experiential models just adapt some especial SAR clutter, and their effect become poorer while the clutter's heterogeneity increases.The paper weightily analyses generalized compound(GC) pdf model which is based on product model. Some special cases got from structure of the model, and the gradual changes of model parameters are described. The research indicated that how to estimate model parameters is an optimization problem, and then a method is brought to search best parameters which combines Quasi-Newton algorithm and random search algorithm. And more, the correctness and applicability are validated by experiments.At last, The paper brings a classification method of SAR image based on GC pdf model. In detail, employing K-neighbour method to collect samples and bayes rule to classify SAR image. The validity is approved by experiments.