With global warming proceeding,sea ice in Northwest Passage has been changing and has increasingly become the focus of attention.Ninety seven(97)Sentinel-1HV polarization images in Northwest Passage from November 2015-February 2018 were used to perform boundary noise removal,angle effect correction and other preprocessing procedures.Grey Level Co-occurrence Matrix method was adopted to extract textural features in SAR imagery and train samples of first-year ice and multi-year ice textural features. Classification was performed using LibSVM.Canadian ice service ice charts were regarded as real sea ice classification data source to assess our classification.The overall classification accuracy was found to be 73.15%.