Research And Application Of Sea Ice Detection Technology Based On Decision Tree Algorithm For Remote Sensing Image

In the recent years, marine disasters have occurred frequently. As China’s northern sea area located in the high latitudes, the sea ice disasters happen sometimes, which makes it an important research fields on Bohai’s environment, resource and disaster to acquire the distribution information of the sea ice. It is of great significance to detect the range of the sea ice distribution, which will contribute to study on the sea ice freezing law and predict the scope of the sea ice, thereby reducing the loss caused by the sea ice disaster. Essentially the detection of the sea ice can be considered as a classification of the sea water and the sea ice into two classes. The differences between the spectral characteristics of the sea ice and the seawater provide the basis for the classification and these spectral data can be provided from Remote Sensing images. As an important data mining technology, the decision tree algorithm can effectively complete the classification task and find the critical factors that will affect the classification process. This paper studies the decision tree-based remote sensing images sea ice detection methods, and a sea ice detection system is built. The main work in this paper is as follows:The remote sensing images-based sea ice detection is transformed into the classification of the sea water and the sea ice. Based on the decision tree algorithm of data mining, the method and its process are given. Through the statistics of the bands of MODIS remote sensing image, the key bands in the MODIS image associated with the sea water and the sea ice are found. Based on this, the relative bands are used to build a decision tree for sea ice detection to acquire the sea ice detection rules, and then the rules are used to detect the sea ice in the Liaodong Bay of Bohai, with the result showing the effectiveness of this method.Based on the study of sea ice detection method, the decision tree-based remote sensing image sea ice detection system is designed and implemented. This system contains a data import module, the data preprocessing module, the information extraction module, and the data output module.