お知らせ:
2025 年 4 月 15 日より前に Earth Engine の使用を登録したすべての非商用プロジェクトは、Earth Engine へのアクセスを維持するために
非商用目的での利用資格を確認する必要があります。
ee.Image.sampleRegions
コレクションでコンテンツを整理
必要に応じて、コンテンツの保存と分類を行います。
1 つ以上のリージョンと交差する画像(特定のスケール)の各ピクセルを Feature に変換し、FeatureCollection として返します。各出力フィーチャーには、入力画像のバンドごとに 1 つのプロパティと、入力フィーチャーからコピーされた指定のプロパティが含まれます。
ジオメトリはピクセルの中心にスナップされます。
用途 | 戻り値 |
---|
Image.sampleRegions(collection, properties, scale, projection, tileScale, geometries) | FeatureCollection |
引数 | タイプ | 詳細 |
---|
これ: image | 画像 | サンプリングする画像。 |
collection | FeatureCollection | サンプリングするリージョン。 |
properties | リスト、デフォルト: null | 各入力特徴からコピーするプロパティのリスト。デフォルトはすべてのシステム以外のプロパティです。 |
scale | 浮動小数点数、デフォルト: null | サンプリングする投影のメートル単位の名義尺度。指定しない場合、イメージの最初のバンドのスケールが使用されます。 |
projection | Projection、デフォルト: null | サンプリングする投影。指定しない場合、イメージの最初のバンドの投影が使用されます。スケールに加えて指定された場合、指定されたスケールに再スケーリングされます。 |
tileScale | 浮動小数点数、デフォルト: 1 | 集計タイルのサイズを縮小するために使用されるスケーリング ファクタ。tileScale を大きくすると(例: 2 または 4)を使用すると、デフォルトでメモリ不足になる計算が可能になる場合があります。 |
geometries | ブール値。デフォルト値は false です。 | true の場合、結果にはサンプリングされたピクセルごとにポイント ジオメトリが含まれます。それ以外の場合、ジオメトリは省略されます(メモリが節約されます)。 |
例
コードエディタ(JavaScript)
// A Sentinel-2 surface reflectance image.
var img = ee.Image('COPERNICUS/S2_SR/20210109T185751_20210109T185931_T10SEG');
Map.setCenter(-122.503881, 37.765588, 18);
Map.addLayer(img, {bands: ['B11', 'B8', 'B3'], min: 100, max: 4500}, 'img');
// A feature collection with two polygon regions each intersecting 36
// pixels at 10 m scale.
var fcPolygon = ee.FeatureCollection([
ee.Feature(ee.Geometry.Rectangle(
-122.50620929, 37.76502806, -122.50552264, 37.76556663), {id: 0}),
ee.Feature(ee.Geometry.Rectangle(
-122.50530270, 37.76565568, -122.50460533, 37.76619425), {id: 1})
]);
Map.addLayer(fcPolygon, {color: 'yellow'}, 'fcPolygon');
var fcPolygonSamp = img.sampleRegions({
collection: fcPolygon,
scale: 10,
geometries: true
});
// Note that 7 pixels are missing from the sample. If a pixel contains a masked
// band value it will be excluded from the sample. In this case, the TCI_B band
// is masked for each unsampled pixel.
print('A feature per pixel (at given scale) in each region', fcPolygonSamp);
Map.addLayer(fcPolygonSamp, {color: 'purple'}, 'fcPolygonSamp');
// A feature collection with two points intersecting two different pixels.
// This example is included to show the behavior for point geometries. In
// practice, if the feature collection is all points, ee.Image.reduceRegions
// should be used instead to save memory.
var fcPoint = ee.FeatureCollection([
ee.Feature(ee.Geometry.Point([-122.50309256, 37.76605006]), {id: 0}),
ee.Feature(ee.Geometry.Point([-122.50344661, 37.76560903]), {id: 1})
]);
Map.addLayer(fcPoint, {color: 'cyan'}, 'fcPoint');
var fcPointSamp = img.sampleRegions({
collection: fcPoint,
scale: 10
});
print('A feature per point', fcPointSamp);
Python の設定
Python API とインタラクティブな開発での geemap
の使用については、
Python 環境のページをご覧ください。
import ee
import geemap.core as geemap
Colab(Python)
# A Sentinel-2 surface reflectance image.
img = ee.Image('COPERNICUS/S2_SR/20210109T185751_20210109T185931_T10SEG')
m = geemap.Map()
m.set_center(-122.503881, 37.765588, 18)
m.add_layer(
img, {'bands': ['B11', 'B8', 'B3'], 'min': 100, 'max': 4500}, 'img'
)
display(m)
# A feature collection with two polygon regions each intersecting 36
# pixels at 10 m scale.
fc_polygon = ee.FeatureCollection([
ee.Feature(
ee.Geometry.Rectangle(
-122.50620929, 37.76502806, -122.50552264, 37.76556663
),
{'id': 0},
),
ee.Feature(
ee.Geometry.Rectangle(
-122.50530270, 37.76565568, -122.50460533, 37.76619425
),
{'id': 1},
),
])
m.add_layer(fc_polygon, {'color': 'yellow'}, 'fc_polygon')
fc_polygon_samp = img.sampleRegions(
collection=fc_polygon, scale=10, geometries=True
)
# Note that 7 pixels are missing from the sample. If a pixel contains a masked
# band value it will be excluded from the sample. In this case, the TCI_B band
# is masked for each unsampled pixel.
display('A feature per pixel (at given scale) in each region', fc_polygon_samp)
m.add_layer(fc_polygon_samp, {'color': 'purple'}, 'fc_polygon_samp')
# A feature collection with two points intersecting two different pixels.
# This example is included to show the behavior for point geometries. In
# practice, if the feature collection is all points, ee.Image.reduceRegions
# should be used instead to save memory.
fc_point = ee.FeatureCollection([
ee.Feature(ee.Geometry.Point([-122.50309256, 37.76605006]), {'id': 0}),
ee.Feature(ee.Geometry.Point([-122.50344661, 37.76560903]), {'id': 1}),
])
m.add_layer(fc_point, {'color': 'cyan'}, 'fc_point')
fc_point_samp = img.sampleRegions(collection=fc_point, scale=10)
display('A feature per point', fc_point_samp)
特に記載のない限り、このページのコンテンツはクリエイティブ・コモンズの表示 4.0 ライセンスにより使用許諾されます。コードサンプルは Apache 2.0 ライセンスにより使用許諾されます。詳しくは、Google Developers サイトのポリシーをご覧ください。Java は Oracle および関連会社の登録商標です。
最終更新日 2025-07-26 UTC。
[null,null,["最終更新日 2025-07-26 UTC。"],[[["\u003cp\u003e\u003ccode\u003eImage.sampleRegions\u003c/code\u003e extracts pixel values from an image within specified regions, returning them as a FeatureCollection.\u003c/p\u003e\n"],["\u003cp\u003eEach output feature contains the band values of the input image for each sampled pixel, along with properties from the input features.\u003c/p\u003e\n"],["\u003cp\u003eGeometries are snapped to pixel centers, and you can control the sampling scale and projection.\u003c/p\u003e\n"],["\u003cp\u003eThe function provides options to include point geometries and adjust the tile scale for memory management.\u003c/p\u003e\n"],["\u003cp\u003eIf the input is a FeatureCollection of points, \u003ccode\u003eee.Image.reduceRegions\u003c/code\u003e is generally recommended for better memory efficiency.\u003c/p\u003e\n"]]],["The `Image.sampleRegions` method converts image pixels intersecting specified regions into a `FeatureCollection`. Each output feature contains properties from the input image bands and any designated input feature properties. Geometries are snapped to pixel centers. The sampling scale and projection can be specified; otherwise, the image's first band defaults are used. Optionally, geometries of the sampled pixels can be included, and tile scaling can be used for memory management.\n"],null,["# ee.Image.sampleRegions\n\nConverts each pixel of an image (at a given scale) that intersects one or more regions to a Feature, returning them as a FeatureCollection. Each output feature will have one property per band of the input image, as well as any specified properties copied from the input feature.\n\n\u003cbr /\u003e\n\nNote that geometries will be snapped to pixel centers.\n\n| Usage | Returns |\n|-----------------------------------------------------------------------------------------------------------------|-------------------|\n| Image.sampleRegions`(collection, `*properties* `, `*scale* `, `*projection* `, `*tileScale* `, `*geometries*`)` | FeatureCollection |\n\n| Argument | Type | Details |\n|---------------|---------------------------|--------------------------------------------------------------------------------------------------------------------------------------------------------------------------|\n| this: `image` | Image | The image to sample. |\n| `collection` | FeatureCollection | The regions to sample over. |\n| `properties` | List, default: null | The list of properties to copy from each input feature. Defaults to all non-system properties. |\n| `scale` | Float, default: null | A nominal scale in meters of the projection to sample in. If unspecified, the scale of the image's first band is used. |\n| `projection` | Projection, default: null | The projection in which to sample. If unspecified, the projection of the image's first band is used. If specified in addition to scale, rescaled to the specified scale. |\n| `tileScale` | Float, default: 1 | A scaling factor used to reduce aggregation tile size; using a larger tileScale (e.g., 2 or 4) may enable computations that run out of memory with the default. |\n| `geometries` | Boolean, default: false | If true, the results will include a point geometry per sampled pixel. Otherwise, geometries will be omitted (saving memory). |\n\nExamples\n--------\n\n### Code Editor (JavaScript)\n\n```javascript\n// A Sentinel-2 surface reflectance image.\nvar img = ee.Image('COPERNICUS/S2_SR/20210109T185751_20210109T185931_T10SEG');\nMap.setCenter(-122.503881, 37.765588, 18);\nMap.addLayer(img, {bands: ['B11', 'B8', 'B3'], min: 100, max: 4500}, 'img');\n\n// A feature collection with two polygon regions each intersecting 36\n// pixels at 10 m scale.\nvar fcPolygon = ee.FeatureCollection([\n ee.Feature(ee.Geometry.Rectangle(\n -122.50620929, 37.76502806, -122.50552264, 37.76556663), {id: 0}),\n ee.Feature(ee.Geometry.Rectangle(\n -122.50530270, 37.76565568, -122.50460533, 37.76619425), {id: 1})\n]);\nMap.addLayer(fcPolygon, {color: 'yellow'}, 'fcPolygon');\n\nvar fcPolygonSamp = img.sampleRegions({\n collection: fcPolygon,\n scale: 10,\n geometries: true\n});\n// Note that 7 pixels are missing from the sample. If a pixel contains a masked\n// band value it will be excluded from the sample. In this case, the TCI_B band\n// is masked for each unsampled pixel.\nprint('A feature per pixel (at given scale) in each region', fcPolygonSamp);\nMap.addLayer(fcPolygonSamp, {color: 'purple'}, 'fcPolygonSamp');\n\n// A feature collection with two points intersecting two different pixels.\n// This example is included to show the behavior for point geometries. In\n// practice, if the feature collection is all points, ee.Image.reduceRegions\n// should be used instead to save memory.\nvar fcPoint = ee.FeatureCollection([\n ee.Feature(ee.Geometry.Point([-122.50309256, 37.76605006]), {id: 0}),\n ee.Feature(ee.Geometry.Point([-122.50344661, 37.76560903]), {id: 1})\n]);\nMap.addLayer(fcPoint, {color: 'cyan'}, 'fcPoint');\n\nvar fcPointSamp = img.sampleRegions({\n collection: fcPoint,\n scale: 10\n});\nprint('A feature per point', fcPointSamp);\n```\nPython setup\n\nSee the [Python Environment](/earth-engine/guides/python_install) page for information on the Python API and using\n`geemap` for interactive development. \n\n```python\nimport ee\nimport geemap.core as geemap\n```\n\n### Colab (Python)\n\n```python\n# A Sentinel-2 surface reflectance image.\nimg = ee.Image('COPERNICUS/S2_SR/20210109T185751_20210109T185931_T10SEG')\nm = geemap.Map()\nm.set_center(-122.503881, 37.765588, 18)\nm.add_layer(\n img, {'bands': ['B11', 'B8', 'B3'], 'min': 100, 'max': 4500}, 'img'\n)\ndisplay(m)\n\n# A feature collection with two polygon regions each intersecting 36\n# pixels at 10 m scale.\nfc_polygon = ee.FeatureCollection([\n ee.Feature(\n ee.Geometry.Rectangle(\n -122.50620929, 37.76502806, -122.50552264, 37.76556663\n ),\n {'id': 0},\n ),\n ee.Feature(\n ee.Geometry.Rectangle(\n -122.50530270, 37.76565568, -122.50460533, 37.76619425\n ),\n {'id': 1},\n ),\n])\nm.add_layer(fc_polygon, {'color': 'yellow'}, 'fc_polygon')\n\nfc_polygon_samp = img.sampleRegions(\n collection=fc_polygon, scale=10, geometries=True\n)\n# Note that 7 pixels are missing from the sample. If a pixel contains a masked\n# band value it will be excluded from the sample. In this case, the TCI_B band\n# is masked for each unsampled pixel.\ndisplay('A feature per pixel (at given scale) in each region', fc_polygon_samp)\nm.add_layer(fc_polygon_samp, {'color': 'purple'}, 'fc_polygon_samp')\n\n# A feature collection with two points intersecting two different pixels.\n# This example is included to show the behavior for point geometries. In\n# practice, if the feature collection is all points, ee.Image.reduceRegions\n# should be used instead to save memory.\nfc_point = ee.FeatureCollection([\n ee.Feature(ee.Geometry.Point([-122.50309256, 37.76605006]), {'id': 0}),\n ee.Feature(ee.Geometry.Point([-122.50344661, 37.76560903]), {'id': 1}),\n])\nm.add_layer(fc_point, {'color': 'cyan'}, 'fc_point')\n\nfc_point_samp = img.sampleRegions(collection=fc_point, scale=10)\ndisplay('A feature per point', fc_point_samp)\n```"]]