Study On Oil Spill Identification Technology Based On Oil Fingerprint Library

With the deepening economic integration and the increasing volume of oil and its products transported by sea,all kinds of oil spill pollution incidents have become more and more frequent.Oil spill disaster is sudden,in the event of marine resources,ecological environment and human health caused great harm.How to detect the oil in time and identify the oil accurately and quickly provides an important basis for the governance of oil spills at sea and the resolution of liability disputes.Adopting machine learning methods for self-adaptive identification of marine oil spills has a broad application prospect.This paper outlines the common methods for detection and identification of oil spills on the sea,researches a method for rapid identification of oil spills on the basis of satellite images,and a method for rapid detection of oil spills on the basis of pattern recognition.This method uses SVM classifiers to implement thermal infrared images based on Rapid detection of oil spills.The specific research content is as follows:(1)Oil Fingerprint Based Satellite Remote Sensing Rapid Identification of Oil Spill on the Sea.Using satellite imagery to study oil spills on the sea,using the maximum likelihood method for classification of processed images,extracting information on oil spilled area and DN values from the images,identifying oil spill types by oil fingerprinting,and estimating oil spills As an example,the “7.16” oil spill in Dalian was used to verify the feasibility of the method.The results showed that the oil spilled from the sea was crude oil,and the oil spill was estimated to be between 13910.63 t and 23184.39 t.(2)Rapid Detection of Oil Spill on the Sea Based on Thermal Infrared Remote Sensing.The use of computer pattern recognition technology for rapid detection of oil spills on the sea,based on the integral channel characteristics and SVM classifier established a marine oil spill detection model to achieve rapid detection of oil spills on the sea,using oil spill image samples to verify the reliability of oil spill detection model The recognition accuracy is 92.25%.