Study On Sea Ice Drift Tracking Based On SAR Images Of Large Difference In Resolution

The freezing and drifting of sea ice has different effects on marine navigation and marine production,even can cause serious disasters every year.Therefore,it is important to carry out the real-time and all-weather tracking of sea ice drift.The sea ice drift tracking with high precision requires high resolution image sequences with short time intervals.As the transit period is long,homologous spaceborne SAR limits the accuracy of sea ice drift tracking.Instead,it is a reasonable solution to study on the sea ice drift tracking with high precision that non homologous SAR is used to construct image sequences with short time intervals.Characteristic correspondence and tracking algorithm efficiency are two important factors that affect the accuracy of sea ice drift tracking.Because of the difference of sensors,the band,polarization,resolution,incidence angle,noise level and so on are different,which leads to obvious characteristic differences between non homologous SAR.In this paper,the study of sea ice drift tracking for non homologous SAR images with large difference of resolution is presented.The main work and research results are summarized as follows:Firstly,the characteristic correspondence of non homologous SAR is focused on.Because the parameters of the same homologous data are same,the corresponding sea ice characteristics are similar.But the difference of parameters between non homologous data lead to the difference in the characteristics of the same sea ice in different data,especially in the situation of large resolution difference,some detail characteristics presenting in high resolution doesn't exist corresponding relationship in low resolution,so this paper puts forward a way by finding the stable characteristics to analyze and research characteristic correspondence based on super resolution reconstruction technology;Secondly,because there are obvious differences in the gray scale between the non homologous data,some classical sea ice drift tracking algorithm(such as MCC algorithm)suitable for homologous data will fail to deal with non homologous data.Therefore,this paper designs and implements an algorithm of characteristic matching and tracking based on the SIFT operator.The calculation of the algorithm is based on structural features rather than the gray features,which can effectively improve the matching and tracking accuracy.What's more,the rotation invariance of SIFT algorithm is the effective solution to the problem of the rotation of sea ice;Finally,because the situation of squeeze and collision will lead that sea ice drift is not a linear change during the process of sea ice drift,this paper uses the classical nonlinear stable interpolation algorithm(ordinary Kriging interpolation),and the initial condition setting of interpolation is improved base on it.Ultimately,the velocity vector field calibration of sea ice drift is realized.