Implementation Of Machine Learning Algorithms For Monitoring And Analyzing Marine Sewage With Remote Sensing Image

For a long time,obtaining data by monitoring the pollution status of Bohai Bay near Tianjin City depends mainly on human chartering.But,this method is time-consuming and labor-intensive.However,the increasing availability of free remote sensing images has enabled us to rely on computers to monitor marine sewage.In this paper,first,the Grab-Cut image segmentation algorithm is used to analyze the pollution situation in Bohai Bay near Tianjin with the US satellite remote sensing images.The conclusion is that the situation of marine sewage in recent years is severe,fortunately,three important pollution outlets of which has been significantly improved since 2015.In addition,this paper compares the improved Grab-Cut algorithm with other common segmentation algorithms.The results show that the Grab-Cut algorithm is superior to the others in segmenting object zone with less user interaction.However,the Grab-Cut algorithm and some computational methods have been proposed nowadays with the image segmentation,which more or less needs the human interaction.We proposed a new method for the marine sewage detection based on a machine learning method,which is Gradient Boosting Decision Tree(GBDT),a boosting method which is fast and highly accurate.Meanwhile,Information Entropy(IFE)is firstly used as a feature in the classification of remote sensing(RS)images.To assess our method,several experiments including feature evaluation,parameter setting and classification were conducted on 100 image cases(50 polluted cases and50 unpolluted cases).The results show that the classification accuracy of our method reaches 98.33%,with the entire process(training and testing)finished in 0.020 s.Our GBDT-based method might therefore aid the marine sewage detection in the future.The pollution monitoring results of the two algorithms are effective and can be used in the future monitoring of marine pollution.