Research On Ocean Enteromorpha Prolifera Monitoring With SAR Data

China is a country with large sea territory, which provides not only abundant tourism, industry and fishery resources, but also important channel of transportation. In recent years, large-scale enteromorpha prolifera blooms outbreak at sea annually with prolonged duration and expending coverage. Such algae blooms serious threaten navigation safety and the coastal environment of Rizhao and Qingdao. Protecting marine environment and ensuring navigation safety has become an important task of the Chinese government. To deal with the problem caused by enteromorpha prolifera bloom, higher monitoring and response capability is required, especially the capability to identify the position of pollution source rapidly and provide early warning information timely. Among the various monitoring technologies, satellite remote sensing, with its large coverage and capability to conduct high-frequency repeated observations, has already become conventional measure of marine pollution monitoring and has been playing increasingly important role.The enteromorpha prolifera bloom is taken as the study object of the article, and various Radarsat-2 images containing information of enteromorpha prolifera blooms are taken as the research data. With combined analysis methodology of feature extraction, data processing and accuracy assessment, this article evaluated the efficiency of information extraction from two traditional supervised classification method, i.e. minimum distance classification and maximum likelihood classification. Based on the analysis of traditional classification methods, this article researched on the object-oriented classification based on decision tree rules, discovered the rules of the decision tree classification model suitable for the extraction of enteromorpha prolifera information, and carried out experiments using decision tree classification algorithm. The major components and conclusions of this article were as follows:Firstly, this article analyzed and tested the traditional image classification technologies, especially on the principles of minimum distance classification and maximum likelihood classification. This article also conducted experimental analysis and accuracy evaluation of different imaging quality. It was concluded that the single image segmentation method was hard to provide satisfying results, and combined segmentation methods should be applied.Secondly, this article researched and experimented on the new method of object-oriented classification based on decision tree rules image classification, with special focus on the principles. This article discovered the rules of the decision tree classification model suitable for the extraction of enteromorpha prolifera information, conducted experimental analysis and the accuracy evaluation, and achieved good results. The experimental results showed that the method of object-oriented classification based on decision tree rules had good classification performance for radar satellite image.Thirdly, the object-oriented classification based on decision tree rules image classification was introduced into the extraction of enteromorpha prolifera information from radar satellite image. Better extraction results were achieved through repeated experiments, and the problems of low accuracy and slow speed in the monitoring of enteromorpha prolifera were solved. This method catered to the need of maritime administrations and significantly improved the efficiency and accuracy of enteromorpha prolifera monitoring.