Study on Superpixel-Level Sea Clutter Statistical Model for SAR Imagery

摘要

With the development of remote sensing technology, the acquired SAR images are gradually developing toward high resolution and large scale. At the same time, the interpretation of remote sensing images has also evolved from the traditional pixel-based processing method to the object-based processing method to meet the need of real-time processing. In this paper, we find that the SciPy statistical function library (SPSFL) can provide more matched distribution models for sea clutters than traditionally used models through experiments on SAR images. Thus, we innovatively using the SPSFL to analyze the distribution model of superpixel-level sea clutters, and the experiment results indicate that the Johnsonsu distribution model fits the superpixel-level sea clutters best under different conditions. In addition, this conclusion is verified by multiple sets of experiments. Research results of this paper can provide theoretical and experimental support for the sea clutters distribution models in superpixel-level remote sensing image interpretation.