GEE22：基于目视解译的土地利用分类（随机森林监督分类）

采样点信息：

设置一下采样点参数：

代码：

//设置研究区位置
var table = ee.FeatureCollection("users/cduthes1991/boundry/China_province_2019");
var roi = table.filter(ee.Filter.eq('provinces','beijing'));
Map.centerObject(roi, 8)
Map.addLayer(roi.style({color:'black',fillColor:'00000000'}),{},'roi');

//设置时间窗口
var startDate = '2023-03-01';
var endDate = '2023-10-01';
var L9 = ee.ImageCollection('LANDSAT/LC09/C02/T1_L2');

// 进行比例换算
function applyScaleFactors(image) {
  var opticalBands = image.select('SR_B.').multiply(0.0000275).add(-0.2);
  var thermalBands = image.select('ST_B.*').multiply(0.00341802).add(149.0);
  return image.addBands(opticalBands, null, true)
              .addBands(thermalBands, null, true);
} 
//去云处理
function rmL89Cloud(image) { 
  var Cirrus = (1 << 2);
  var cloud = (1 << 3); 
  var cloudsShadow = (1 << 4); 
  var qa = image.select('QA_PIXEL'); 
  var mask = qa.bitwiseAnd(Cirrus).eq(0) 
                 .and(qa.bitwiseAnd(cloud).eq(0))
                 .and(qa.bitwiseAnd(cloudsShadow).eq(0)); 
  return image.updateMask(mask); 
}
//指数计算
function addIndexs (image){
     var ndvi = image.normalizedDifference(['SR_B5','SR_B4']).rename('ndvi');
     var bsi = image.expression(
         "((SR_B7+SR_B3)-(SR_B4-SR_B1))/(SR_B7+SR_B3)+(SR_B4-SR_B1)",
          {
            "SR_B7": image.select('SR_B7'),
            "SR_B4": image.select('SR_B4'),
            "SR_B3": image.select('SR_B3'),
            "SR_B1": image.select('SR_B1')
          }
      ).rename('bsi');
      var lswi = image.normalizedDifference(["SR_B4", "SR_B7"]).rename('lswi');
      var mndwi = image.addBands(image.normalizedDifference(["SR_B1", "SR_B7"]).rename('mndwi'));
      return image.addBands(ndvi).addBands(bsi).addBands(lswi).addBands(mndwi);
}
//获取研究区时间窗口内数据
var img = L9.filterDate(startDate,endDate)
            .filterBounds(roi)
            .map(applyScaleFactors)
            .map(addIndexs)
            .map(rmL89Cloud)
            .median()
            .clip(roi)
            .select(['SR_B2','SR_B3','SR_B4','SR_B5','SR_B6','SR_B7','ndvi','bsi','lswi','mndwi'])
var visualization = {
  bands: ['SR_B4', 'SR_B3', 'SR_B2'],
  min: 0.0,
  max: 0.3,
};
Map.addLayer(img,visualization,'img');
print(img)

//合并样本
var samples = water.merge(building).merge(forest);
print('Sample Collection', samples)

//利用样本点对遥感数据进行采样
var samples_img = img.sampleRegions({
  collection:samples,
  properties:['landcover'],
  scale:30,
  tileScale:8,
  geometries:true,
})
print("sample information: ", samples_img)

//给样本添加随机数，用于训练和验证
var samples_img = samples_img.randomColumn();
print('samples_img',samples_img)
// 将样本随机分为7:3，70%用于训练，30%用于验证
var training_samples = samples_img.filter(ee.Filter.lte('random',0.7));
var validation_samples = samples_img.filter(ee.Filter.gte('random',0.7));
//设置分类器
var classifer = ee.Classifier.smileRandomForest(300).train({
  features:training_samples,
  classProperty:'landcover',
  inputProperties:img.bandNames(),
})
//利用分类器对影像进行分类
var Classified_RF = img.classify(classifer)
var visParam = [
  'blue', // 水体water
  'red', // 建筑用地urban
  'green', // 林地forest
];

//显示分类结果（色彩设置依据IGBG）
Map.addLayer(Classified_RF,{'palette':visParam, min:1,max:3},'Classified_RF');
//精度验证
var test = validation_samples.classify(classifer);
var confussionMatrix = test.errorMatrix('landcover','classification');
var accuracy = confussionMatrix.accuracy();
var kappa = confussionMatrix.kappa();
var userAccuracy = confussionMatrix.consumersAccuracy();
var productAccuracy = confussionMatrix.producersAccuracy();
print('accuracy',accuracy);
print('kappa',kappa);
print('userAccuracy',userAccuracy);
print('productAccuracy',productAccuracy);

// 图像导出
Export.image.toDrive({
    image: Classified_RF,
    description: 'Classified_RF',
    region: roi,
    scale: 250,
    maxPixels: 1e13,
	folder: 'LUCC'
})

结果展示：

以上结果只是一个简单示例，需要设置更多的采样点才能更精确分类。

• GEE代码链接