2. traObjExp

/*  Author: Afag Rizayeva
    Date: 01-17-2022
    Purpose: Export the objects overlapping with point grid for further labeling
    Inputs: 2 Corona image layers manually imported in GEE
            Point grid
    
    Adjust the input files' directories
    Run the code
    Export the accuracies as csv files and the final classification map
*/


//     Import two Corona images
//     Rename the band names for afterward- ('da') and forward-facing ('df') cameras
var image2 = ee.Image('users/rizayeva/corona_2_5m/DS1011-1040DA_2_5m_JPEG_1band_EPSG32638').rename(['da']);
var image3 = ee.Image('users/rizayeva/corona_2_5m/DS1011-1040DF_2_5m_JPEG_1band_EPSG32638').rename(['df']);

//     Concatenate the two images
var img = ee.Image.cat([image2, image3]);
print(img);
// print('Band names: ', img.bandNames()); // ee.List of band names

var img = ee.Image(img).divide(255);
// Map.addLayer(img, {bands: ["df", "df", "da"], gamma: 0.8}, 'DS_DADF');

//     Center the map view 
Map.centerObject(img.geometry(), 7);

//     Import and visualize the point grid for extraction of the segment grid
var table4 = ee.FeatureCollection('users/rizayeva/grid5km_label_07142020_EPSG32638');
Map.addLayer(table4, {}, 'training points');

///////////// Run segmentation

//     Set the seed grid size
var seeds = ee.Algorithms.Image.Segmentation.seedGrid(40);

//     Run SNIC segmentation algorithm on the regular square grid
var snic = ee.Algorithms.Image.Segmentation.SNIC({
  image: img, 
  size: 32,
  compactness: 0.1,
  connectivity: 4,
  neighborhoodSize:256,
  seeds: seeds
}).select(['da_mean', 'df_mean', 'clusters'], ['da', 'df', 'clusters']).reproject({crs: 'EPSG:4326', scale: 18});
// print('Snic band names: ', snic.bandNames()); // ee.List of band names

//     Select clusters from segmentation results
var clusters = snic.select('clusters');
// Map.addLayer(clusters.randomVisualizer(), {}, 'clusters', false)
// Map.addLayer(snic, {bands: ['da', 'df'], min:0, max:1, gamma: 0.8}, 'means', false);

///////////// Vectorize the objects

//     Create a cell grid across the image
var tiles = image2.addBands(image3).reduceToVectors({
  reducer: ee.Reducer.mean(), 
  geometry: image2.addBands(image3).geometry(), 
  geometryType: 'bb',
  scale: 12000, 
  eightConnected: false,
  labelProperty: 'object',
});
Map.addLayer(tiles, {}, 'tiles2', false);

///////////// Create buffers for the cell grid across the image to prevent clipped segments later on

//     Build the function
var buffer = function(feature) {
return feature.buffer(1000);};
//     Map the function
var tilesbuffer = tiles.map(buffer);
// print('tilesbuffer', tilesbuffer);

//     Get the total number of the cells across the image
var tilesbuffer_list =tilesbuffer.toList(tilesbuffer.size());
// print('tilesbuffer_list', tilesbuffer_list);
Map.addLayer(tilesbuffer, {}, 'tilesbuffer');
var size = tilesbuffer_list.size().getInfo();
// ee.Feature(collection.toList(3).get(1))

//     Select the column from the training data that has unique values for each of the points in the point grid
var classes = ee.FeatureCollection(table4.select('FID_test_l'));
print('classes', classes.first());


///////////// Get segments overlapping with the point grid

//     Define a spatial filter as geometries that intersect.
var spatialFilter = ee.Filter.intersects({
  leftField: '.geo',
  rightField: '.geo',
  maxError: 10
});

//     Define a save all join.
var saveAllJoin = ee.Join.saveAll({
  matchesKey: 'FID_test_l',
});

//     Create a for loop to extract all segments that overlap with the point grid
var trainobjs = [];
var i;
for (i = 0; i < size; i++) {
var trainingobjects = saveAllJoin.apply(clusters.reduceToVectors({
  reducer: ee.Reducer.countEvery(), 
  geometry: ee.Feature(tilesbuffer_list.get(i)).geometry(),
  scale: 2.5,
  eightConnected: false,
  maxPixels: 1e15
}), classes, spatialFilter);
trainobjs[i]=trainingobjects;   
}
print('trainobjs', trainobjs);

//     Convert the list of Feature collections into a Feature collection of Feature collections
var trainobjsfeat = ee.FeatureCollection(trainobjs);
print('Feature collection of feature collections - trainobjsfeat', trainobjsfeat);

//     To get a Feature collection, flatten the Feature collection of Feature collections 
var trainobjsfeat = trainobjsfeat.flatten();
print('Feature collection - trainobjsfeat', trainobjsfeat);
Map.addLayer(trainobjsfeat, {}, 'trainobjsfeat', false);

//     Export segment grid that overlaps with the point grid in GeoJSON format
Export.table.toDrive({
  collection: trainobjsfeat,//class,
  description: 'trainobjsfeatclass_DS10',
  fileFormat: 'GeoJSON'//'shp'
});