The Application Of Neural Network For Ice Recognition In The SAR Sea Ice Image

The freezing,melting and driftage of the sea ice will not only influence the global climate, but also will be in the way of sea petrol exploitation and endanger the traffic. To say nothing of the sea ice disaster, in the past years, it has done great damage to the sea traffic, ocean engineering. In recent years, several developed countries are doing research work in using SAR (Synthetic Aperture Radar) to evaluate the ice forecast accuracy and quality. Their research work included the arithmetic for ice recognition, ice classification, and automatic ice monitoring systems, etc.The economy situation in pan-BoHai region is very good these years. But the biggest problem is the ice freezing every winter. In order to solve this problem, our country decides to advance the research work about using SAR on ice monitoring and forecast. By now, our country has some hardware foundation, but has no related software in intelligent processing on ice information. And the key of the processing is the pattern recognition arithmetic.In This article, the author propose a method of using ANN(Artificial Neural Network) incorporated with SAR image texture features to solve ice recognition problems of SAR image. Since ANN is adapted to solve non-linear problems, and the texture features analyzing method is the most popular methods in image processing fields, the author combines these two methods in ice recognition problems, take the texture features as the input of the ANN, and output the recognition results. The results of the experiments show that this method is an effective way in ice recognition of SAR image. And also show that ANN has a brilliant feature in this field.