Sea Ice Classification And Trafficability Analysis In The Northwest Passage Using COSMO-SkyMed High Resolution SAR Images

The Northwest Passage,as the shortest route connecting the Atlantic and Pacific Oceans,has significant economic and strategic implications.The widespread distribution of sea ice is a distinct feature of the area and has seriously affected the safety of navigation.In order to ensure the safety of navigation in the ice area,it is necessary to analyze its accessibility.Therefore,it is necessary to master a certain degree of sea ice information,such as sea ice type,sea ice concentration,etc.Generally speaking,when the sea ice concentration is less than 60%,You can sail.At present,a variety of tools and instruments are used for sea ice monitoring,such as buoys,icebreakers,airplanes,etc.,but such data sources are difficult to perform realtime,large-scale sea ice observations,so they are used for a wide range of areas such as the Northwest Passage.It is a satellite remote sensing technology,and synthetic aperture radar(SAR)is widely used in sea ice monitoring because it is weather-proof and can acquire data all day,all weather and high resolution.The means of sea ice monitoring are also handled by manual interpretation,and the resolution of multi-use data is generally low,and the classification efficiency is not high.It is difficult to describe the details of sea ice,which may cause misclassification.In view of this situation,this paper uses COSMO-SkyMed high-resolution SAR data to classify the sea ice and seawater of the target image;use the classification result to calculate the sea ice concentration of the target area;and combine the channel width to realize the passability analysis of the channel area..Contrast was also performed using Sentinel-1 data,which is widely used in polar research.1.Screen out the Barrow Channel area of the Northwest Passage as an experimental area,using COSMO-SkyMed's Level 1C product(GEC)HH polarized spotlight mode(Spotlight)data and the pre-processed Sentinel-1 data level 1 Ultra-wide mode data of HH polarization from image(GRD)products;2.Using superpixel segmentation to improve classification efficiency;3.Extracting contrast,energy,homogeneity and correlation features in gray level co-occurrence matrix(GLCM)The combination feature of grayscale features is used as the feature of sea ice image to construct the sample set.4.The support vector machine(SVM)model is used to classify the training samples to obtain the classifier model,and the unsupervised classification of sea ice is realized.5.Using the results of sea ice classification,the sea ice intensity calculation is carried out on the experimental sea area to analyze the passability of the experimental area.The specific method is to delineate the sliding window according to the classification result and the course width required by the ship mass.The sea ice concentration of the window is determined to be an obstacle based on the ant colony algorithm above 60%,thereby realizing the passability analysis of the channel area.Experiments show that the proposed feature of the proposed method improves the classification accuracy compared to the single use of GLCM or grayscale features.When the grayscale feature weight of the combined feature is 5,the method of constructing samples using the combined features satisfies the sea ice classification.The requirements of the classification can be used to calculate the sea ice concentration in the ice area;the COSMO-SkyMed data shows better applicability in the specific area of the passability problem.