Multi-resource Data-based Research On Remote Sensing Extraction Algorithm Over The Green Tide In The Yellow Sea

In recent years,there are different level green tide disasters have occurred in many offshore areas from Dalian in the north to Sanya in the south of China.Especially in 2008,the outbreak of Enteromorpha in Qingdao seriously affected the smooth progress of the 29 th Olympic Sailing Regatta.Because the presence of mixed pixels in the traditional low and medium resolution remote sensing images,the extraction area of the green tide is much higher than that of the true value.Green tide extraction based on SAR is still dominated by the single-band threshold method,but this method is unstable,affected by human factors and less effective.There are some disadvantages such as the green tide range extracted by the same threshold value of the green tide aggregation area is larger than the actual area,the scattered distribution area is less than the actual coverage area,etc.,The main research of this paper is as follows:(1)In view of the low resolution images to extract green tide covers a larger area,resolution of airborne SAR data respectively by 3 m,16 m resolution high score 1 satellite data as a benchmark,to improve and establish the green tide of MODIS and GOCI area covered by fine extraction model.The improved model can be used to solve the problem of the large area of green tide covered by low resolution(2)Based on GOCI data,the cloud and fog extraction algorithm was used to count the cloud cover of the green-tide-prone areas of the Yellow Sea from 2015 to 2016.It was concluded that the Yellow Sea area was seriously affected by cloud and fog in the summer.The available data obtained by traditional optical influence is less,and it is necessary to monitor the yellow sea area with the joint SAR data to ensure the continuous monitoring of the green tide disaster;Based on the GOCI images available in 2015,the spatial and temporal variation of the outbreak of the green tide disaster in 2015 was analyzed.Executing multi-type polarization decomposition based on full-polarization SAR image data,the current mainstream target polarization decomposition method is applied to the classification of green tide.Full polarized image are simultaneously applied to the classification of green-wave polarimetric images,and the most useful polarization parameters for classification are selected by the feature selection method.(3)For the full-polarization image data,a corresponding comparison was made between the polarization data of different polarization methods for the detection ability of green tide.The results show that VV polarization is more suitable for the operational monitoring of green tide than other three polarization methods.The green tide classified and extracted Based on the improved Level-Set method and the polynomial regression fitting algorithm show that The green tide extraction based on the Level-Set algorithm and the polynomial regression fitting algorithm is basically consistent with that based on the traditional SAR green tide extraction algorithm,but the accuracy is better than the traditional single-band threshold method,and the calculation speed is faster,so it is suitable for rapid batch processing,It provides a reference for the rapid extraction of green tide disasters.