- Based on YuanhaoGong/SideWindowFilter (https://github.com/YuanhaoGong/SideWindowFilter) open source.
- Adding HSI datasets and some filtering code ( all data were classified using SVM. Some datasets are too large and cannot be uploaded to the GitHub ).
The data sets and number of training samples can be selected within the code.
- DataSetName in {'Indianpines', 'PaviaU', 'XuZhou'}
- classifier in {'raw', 'lgs_box'}
- train_number indicates the number of training samples selected for each category.
- PC indicates the dimension after PCA.
For example, when DataSet Indianpines takes 5-30 training sample points for each class, the method of calling lgs_box algorithm for 10 iterations is as follows:
- demo('Indianpines', 'lgs_box', [5,10,15,20,25,30])
Then the results are then automatically saved to file "all_result.txt".