// 研究区域

var roi = ee.FeatureCollection("users/comingboy1118/China/CH_shi");

var roi = roi.filter(ee.Filter.eq("市","武汉市"));

Map.addLayer(roi,{"color":'red',},"roi");

Map.centerObject(roi, 8);

// 获取影像

var image = ee.ImageCollection("LANDSAT/LC08/C01/T1_SR")

.filterBounds(roi)

.filterDate("2019-01-01","2019-12-31")

.median()

.clip(roi);

// 选取某些波段用于分类

var bands = ['B2', 'B3', 'B4', 'B5', 'B6', 'B7', 'B10', 'B11'];

var image = image.select(bands);

Map.addLayer(image, {bands: ['B4', 'B3','B2'],min: 0, max: 10000,gamma:3,opacity:1},'image');

// 合并训练的数据（这里使用面，不是用点，为了更快的选取训练数据，点都是一个一个的选取，但是我画一个面就有很多个点数据）

var polygon =ee.FeatureCollection([water,build,other]);

// 通过选取的面，转为点数据，实际其实是根据选取的位置获取图像上的特征

var points = image.sampleRegions({

collection: polygon,

properties: ['class'],

scale: 30

});

// 以下就是纯粹为了选取XXX个点的数据进行训练模型，XXX个点数据进行验证模型

// 备注：有一个bug,如果点的数据少于输入的数据，那这个函数就没有意义了

// split train and test points

var split = function(data,number){

var tr = data.randomColumn('random', 555) ;

var train = ee.FeatureCollection([]);

var test = ee.FeatureCollection([]);

for(var i =1;i<=3;i++){

var cate = tr.filter(ee.Filter.eq("class",i)).limit(number,"random");

var cate_tr = cate.limit(number*0.8,"random",false);

var cate_test = cate.limit(number*0.2,"random",true);

train = train.merge(cate_tr);

test = test.merge(cate_test);

}

return [train,test];

};

var train_test = split(points,200);

var train = train_test[0];

var test = train_test[1];

// 统计每类有多少个数据

var figure_histogram = function(points){

var histogram = ui.Chart.feature.histogram({

features: points,

property: "class",

minBucketWidth:1,

maxBuckets:3,

}).setOptions({

legend: {'position': 'none' },

title: 'counts of different categories',

hAxis: {'title': 'categories'},

vAxis: {'title': 'counts'},

});

print(histogram)

return histogram

}

// 训练数据

figure_histogram(train);

// 验证数据

figure_histogram(test);

// 用默认的参数训练SVM模型

var trained = ee.Classifier.libsvm().train(train,'class',bands)

// 对图像进行分类

var classified = image.select(bands).classify(trained);

Map.addLayer(classified, {min: 1, max: 3, palette: ['blue', 'red','green']}, 'classified');

// 验证模型

var validated = test.classify(trained);

var testAccuracy = validated.errorMatrix('class', 'classification');

var accuracy = testAccuracy.accuracy();

var userAccuracy = testAccuracy.consumersAccuracy();

var producersAccuracy = testAccuracy.producersAccuracy();

var kappa = testAccuracy.kappa();

print('Validation error matrix: ', testAccuracy);

print('Validation overall accuracy: ', accuracy);

print('User acc: ', userAccuracy);

print('Prod acc: ', producersAccuracy);

print('Kappa: ', kappa);