Study On Sea Ice Drift Tracking Based On SAR Images Of C And L Bands

In the study of sea ice drift tracking,it is difficult to obtain the homologous SAR image sequence in short time intervals because of the limitation of SAR satellite transit time.So the combination of non homologous SAR data to research the tracking can be considered.However,non homologous SAR image tend to have different bands.Affected by SAR inherent imaging principle,for C and L bands in the same position,the ice type information they show is different.Because the ice type information is not the same,it will affect the accuracy of sea ice tracking.In this paper,based on the difference of non homologous data,a new strategy of sea ice drift tracking based on feature consistency analysis is proposed.The strategy is that the single feature is selected by the corresponding consistency analysis method,and the sea ice drift tracking is carried out in the corresponding region of the multi-feature fusion classification.The main work and research results are summarized as follows:First of all,according to the feature differences of sea ice image from non homologous SAR,a new method for analyzing the corresponding consistency of feature selection is proposed.According to calculating the consistency rate of ice types after classification and fitting the surface of Three-dimensional Probability Density Function(3-D PDF),the suitable single texture features is selected by the consistency rate of ice types and fitting the surface of 3-D PDF.Secondly,a kind of sea ice drift tracking strategy based on the consistent classification region sampling is proposed.According to the prior knowledge of remote sensing image,the classification based on multi-feature fusion is better than single feature,so based on method of multi-objective fuzzy decision,the selected single features are fused to classify sea ice,then the favorable areas of classification are obtained by comparing classification results.The normalized cross correlation algorithm(Normalized Cross-correlation)is used to study on sea ice drift tracking and draw the sea ice drift velocity vector field by Kriging Interpolation.Finally,a method is presented to verify the reliability of the sea ice drift velocity field.Selecting two groups of samples in the preselected area to track and setting different weight of the offset angle and offset distance,the weight of the difference of the two velocity field can be calculated.The smaller the weight is,the smaller the difference is,and the more reliable the velocity field is.A reliable ice type is selected to study by comparing the difference of velocity fields between different ice types.Sample tracking areas are selected from the area of corresponding consistent classification and inconsistent classification.By comparing differences between the generated velocity field,it is verified that selecting sampling area for tracking can improve the tracking accuracy in the area of corresponding consistent classification.