GEE中对哨兵1(Sentinel-1)数据进行RefinedLee滤波处理

var roi = table;
var start = '2022-05-01';
var end = '2022-06-30';
Map.centerObject(roi, 10);

var S1_COL = ee.ImageCollection('COPERNICUS/S1_GRD')
    .filterBounds(roi)
    .filterDate(start, end)
    .filter(ee.Filter.listContains('transmitterReceiverPolarisation', 'VH'))
    .filter(ee.Filter.listContains('transmitterReceiverPolarisation', 'VV'))
    .filter(ee.Filter.eq('instrumentMode', 'IW'))
    .map(function(image){
  return image.select('VH').clip(roi); // 选择 'VV'/'VH' 波段并按研究区域裁剪
});
print('S1_COL:', S1_COL); 

var S1_COL_RL1 = S1_COL.map(refinedLee); //进行RefinedLee滤波
print("S1_COL_RL1:",S1_COL_RL1);

var num = S1_COL_RL1.toList(S1_COL.size());
var list = S1_COL_RL1.getInfo(); // 获取影像信息 
var fileNamePrefix = 'VH_RL';

// 获取影像逐个加载数据在地图
for (var i = 0; i < num.size().getInfo(); i++) {
  var image = ee.Image(num.get(i)).clip(roi);
  var imageID = list.features[i].properties['system:index'];
  var imageVH = image.select('constant');
  // var imageVV = image.select('VV');
  var fileName = fileNamePrefix + '_' + imageID;
  Map.addLayer(imageVH, {min: -35, max: 0}, imageID);
  // Map.addLayer(imageVV, {min: -35, max: 0}, imageID);

  // 导出影像
  Export.image.toDrive({
    image: imageVH,
    description: fileName,
    folder: '2022-DT-S1-VH-RL',
    scale: 10,  
    region: roi.geometry(),
    fileFormat:"GeoTIFF", 
    crs: "EPSG:4326"
  });
}

// powerToDb
function powerToDb(img){
  return ee.Image(10).multiply(img.log10());
}

//dbToPower

function dbToPower(img){
  return ee.Image(10).pow(img.divide(10));
}

// The RefinedLee speckle filter
function refinedLee(image) {
  
  var bandNames = image.bandNames();
  image = dbToPower(image);
  
  var result = ee.ImageCollection(bandNames.map(function(b){
    // var img = image.select([b]);
    var img = image;
    // img must be in natural units, i.e. not in dB!
    // Set up 3x3 kernels 
    var weights3 = ee.List.repeat(ee.List.repeat(1,3),3);
    var kernel3 = ee.Kernel.fixed(3,3, weights3, 1, 1, false);
    var mean3 = img.reduceNeighborhood(ee.Reducer.mean(), kernel3);
    var variance3 = img.reduceNeighborhood(ee.Reducer.variance(), kernel3);
  
    // Use a sample of the 3x3 windows inside a 7x7 windows to determine gradients and directions
    var sample_weights = ee.List([[0,0,0,0,0,0,0], [0,1,0,1,0,1,0],[0,0,0,0,0,0,0], [0,1,0,1,0,1,0], [0,0,0,0,0,0,0], [0,1,0,1,0,1,0],[0,0,0,0,0,0,0]]);
  
    var sample_kernel = ee.Kernel.fixed(7,7, sample_weights, 3,3, false);
  
    // Calculate mean and variance for the sampled windows and store as 9 bands
    var sample_mean = mean3.neighborhoodToBands(sample_kernel); 
    var sample_var = variance3.neighborhoodToBands(sample_kernel);
  
    // Determine the 4 gradients for the sampled windows
    var gradients = sample_mean.select(1).subtract(sample_mean.select(7)).abs();
    gradients = gradients.addBands(sample_mean.select(6).subtract(sample_mean.select(2)).abs());
    gradients = gradients.addBands(sample_mean.select(3).subtract(sample_mean.select(5)).abs());
    gradients = gradients.addBands(sample_mean.select(0).subtract(sample_mean.select(8)).abs());
  
    // And find the maximum gradient amongst gradient bands
    var max_gradient = gradients.reduce(ee.Reducer.max());
  
    // Create a mask for band pixels that are the maximum gradient
    var gradmask = gradients.eq(max_gradient);
  
    // duplicate gradmask bands: each gradient represents 2 directions
    gradmask = gradmask.addBands(gradmask);
  
    // Determine the 8 directions
    var directions = sample_mean.select(1).subtract(sample_mean.select(4))
                                .gt(sample_mean.select(4).subtract(sample_mean.select(7)))
                                .multiply(1);
    directions = directions.addBands(sample_mean.select(6).subtract(sample_mean.select(4))
                          .gt(sample_mean.select(4).subtract(sample_mean.select(2)))
                          .multiply(2));
    directions = directions.addBands(sample_mean.select(3).subtract(sample_mean.select(4))
                          .gt(sample_mean.select(4).subtract(sample_mean.select(5)))
                          .multiply(3));
    directions = directions.addBands(sample_mean.select(0).subtract(sample_mean.select(4))
                          .gt(sample_mean.select(4).subtract(sample_mean.select(8)))
                          .multiply(4));
    // The next 4 are the not() of the previous 4
    directions = directions.addBands(directions.select(0).not().multiply(5));
    directions = directions.addBands(directions.select(1).not().multiply(6));
    directions = directions.addBands(directions.select(2).not().multiply(7));
    directions = directions.addBands(directions.select(3).not().multiply(8));
  
    // Mask all values that are not 1-8
    directions = directions.updateMask(gradmask);
  
    // "collapse" the stack into a singe band image (due to masking, each pixel has just one value (1-8) in it's directional band, and is otherwise masked)
    directions = directions.reduce(ee.Reducer.sum());  
  
    //var pal = ['ffffff','ff0000','ffff00', '00ff00', '00ffff', '0000ff', 'ff00ff', '000000'];
    //Map.addLayer(directions.reduce(ee.Reducer.sum()), {min:1, max:8, palette: pal}, 'Directions', false);
  
    var sample_stats = sample_var.divide(sample_mean.multiply(sample_mean));
  
    // Calculate localNoiseVariance
    var sigmaV = sample_stats.toArray().arraySort().arraySlice(0,0,5).arrayReduce(ee.Reducer.mean(), [0]);
  
    // Set up the 7*7 kernels for directional statistics
    var rect_weights = ee.List.repeat(ee.List.repeat(0,7),3).cat(ee.List.repeat(ee.List.repeat(1,7),4));
  
    var diag_weights = ee.List([[1,0,0,0,0,0,0], [1,1,0,0,0,0,0], [1,1,1,0,0,0,0], 
      [1,1,1,1,0,0,0], [1,1,1,1,1,0,0], [1,1,1,1,1,1,0], [1,1,1,1,1,1,1]]);
  
    var rect_kernel = ee.Kernel.fixed(7,7, rect_weights, 3, 3, false);
    var diag_kernel = ee.Kernel.fixed(7,7, diag_weights, 3, 3, false);
  
    // Create stacks for mean and variance using the original kernels. Mask with relevant direction.
    var dir_mean = img.reduceNeighborhood(ee.Reducer.mean(), rect_kernel).updateMask(directions.eq(1));
    var dir_var = img.reduceNeighborhood(ee.Reducer.variance(), rect_kernel).updateMask(directions.eq(1));
  
    dir_mean = dir_mean.addBands(img.reduceNeighborhood(ee.Reducer.mean(), diag_kernel).updateMask(directions.eq(2)));
    dir_var = dir_var.addBands(img.reduceNeighborhood(ee.Reducer.variance(), diag_kernel).updateMask(directions.eq(2)));
  
    // and add the bands for rotated kernels
    for (var i=1; i<4; i++) {
      dir_mean = dir_mean.addBands(img.reduceNeighborhood(ee.Reducer.mean(), rect_kernel.rotate(i)).updateMask(directions.eq(2*i+1)));
      dir_var = dir_var.addBands(img.reduceNeighborhood(ee.Reducer.variance(), rect_kernel.rotate(i)).updateMask(directions.eq(2*i+1)));
      dir_mean = dir_mean.addBands(img.reduceNeighborhood(ee.Reducer.mean(), diag_kernel.rotate(i)).updateMask(directions.eq(2*i+2)));
      dir_var = dir_var.addBands(img.reduceNeighborhood(ee.Reducer.variance(), diag_kernel.rotate(i)).updateMask(directions.eq(2*i+2)));
    }
  
    // "collapse" the stack into a single band image (due to masking, each pixel has just one value in it's directional band, and is otherwise masked)
    dir_mean = dir_mean.reduce(ee.Reducer.sum());
    dir_var = dir_var.reduce(ee.Reducer.sum());
  
    // A finally generate the filtered value
    var varX = dir_var.subtract(dir_mean.multiply(dir_mean).multiply(sigmaV)).divide(sigmaV.add(1.0));
  
    var b = varX.divide(dir_var);
  
    return dir_mean.add(b.multiply(img.subtract(dir_mean)))
      .arrayProject([0])
      // Get a multi-band image bands.
      .arrayFlatten([['sum']])
      .float();
  })).toBands().rename(bandNames);
  return powerToDb(result);
}

//RL滤波的另一种代码处理：

var s1 = ee.ImageCollection('COPERNICUS/S1_GRD')
    .filterBounds(roi)
    .filterDate(start, end)
    .filter(ee.Filter.listContains('transmitterReceiverPolarisation', 'VH'))
    .filter(ee.Filter.listContains('transmitterReceiverPolarisation', 'VV'))
    .filter(ee.Filter.eq('instrumentMode', 'IW'))
    .map(function(image){
          return image.clip(roi); // 按研究区域裁剪
                  });
print('s1', s1); 
 var S1_RL1_COL = s1.map(function(image) {
var vhImage = image.select('VH'); // 选择 "VH" 波段
var processedVH = refinedLee(vhImage); // 对 "VH" 波段应用 refinedLee1
        return processedVH; // 返回处理后的 "VH" 波段
 });

print("S1_RL1_COL:" ,S1_RL1_COL);

文章为我学习过程的记录，如有侵权，请联系删除。