Title: A new GMRF self-supervised algorithm applied to SAR image classification

Abstract

With the global greenhouse effect, the sea surface coverage is increasing year by year, and people's demand for sea surface monitoring is also increasing, SAR technology is currently an important method for sea ice monitoring, and automatic interpretation of SAR sea ice images has become a key technology in this research field. The problems of limited labeled training samples and unknown the number of classes are challenging for SAR sea ice imagery classification. To handle this,we presents a new GMRF self-supervised algorithm for SAR image. We add a GOF process in the process of estimating GMM parameters by EM algorithm, which can not only dynamically select the best number of significant classes but also provides an initial feature parameter to calculate the MRF minimum energy. After the iterative region label and region growth cycle, iteration is combined with the Mll context model to obtain the best mark of each region. Since the initial feature parameter selection of the MRF is not random, the operation efficiency is also improved while reducing the number of iteration cycles of the algorithm. The experimental results show that this algorithm not only solves the problem of manual input of the number of classes in the unsupervised image classification process, but also provide the better output result graph in terms of detail maintenance than the expert interpretation of the truth map, and we hope that it could support operations and meet the real-time requirements.