| In recent years,with the rapid development of the world economy and society,oil,as one of the most important primary energy sources,has been increasingly demanded by various countries,and the sudden oil spills caused by this have frequently occurred,which has become a particularly serious problem of marine environmental pollution.At present,the most widely used oil spill monitoring method is remote sensing monitoring,which is mainly equipped with three remote sensing equipment platforms: spaceborne,airborne,and shipborne.Among them,the UAV-borne platform has the characteristics of fast,efficient,and real-time and it is the most important means of emergency oil spill remote sensing monitoring.Therefore,the research on remote sensing technology of UAV-borne oil spill is of great significance both at the theoretical level and practical application level.This paper studies the dual-channel oil spill remote sensing monitoring method based on the LIF(laser induced fluorescence)carried by the drone and the visual camera.LIF can effectively detect information such as the type of oil spilled and the thickness of the oil,the camera can effectively detect information such as the area of the oil spill,and the dual-channel coordinated operation can efficiently realize the monitoring task of the oil spill area.First,for the oil spill remote sensing image of aerial photography,this paper proposes a wavelet adaptive threshold-based oil spill image denoising method.Based on the wavelet threshold denoising method,this method adapts the thresholds of different scales of wavelet decomposition.Processing to reduce the noise generated by some pictures affected by the shooting equipment and shooting environment.Secondly,using the SLIC(Simple Linear Iterative Clustering)algorithm,segment the oil spill image into a number of super pixel block regions,and extract the color and texture features of each super pixel block region.Finally,a machine learning algorithm is used to classify the oil spill block and the sea water block to realize the oil-water segmentation of the oil spill image.The experiment is compared with other models on the established data set,and the results show that the method in this paper can identify the oil spill area very well,with high accuracy and practicality.Finally,in order to assess the amount of oil spilled in the oil spill area,this paper designs an oil spill remote sensing monitoring method based on the coordinated operation of LIF and visual cameras.This method first uses Zhang Zhengyou’s calibration method to calibrate the camera’s internal parameters,external parameters,distortion and other parameter information,complete the image correction,and then combine the flying height of the drone to obtain the ground resolution of the oil spill image,thereby calculating and evaluating the target area Area.LIF can efficiently detect the oil spill thickness,while the SLIC algorithm can group the areas with the same oil spill thickness into a super pixel block.The LIF device only needs to detect the oil spill thickness at the centroid point of the super pixel block,and then the overflow of the super pixel block area The oil thickness is the thickness measured at the center of mass point,combined with the calculated oil spill area information,so that the amount of oil spilled in the oil spill area can be evaluated,and the monitoring target of oil spill remote sensing can be effectively realized. |
