Reproject method for ImageData

CHANGES.md

@@ -1,5 +1,16 @@
 # Unreleased (TBD)
 
+* update rasterio dependency to `>=1.4.0`
+
+* add `reproject` method for `ImageData` objects (author @emmanuelmathot, https://github.com/cogeotiff/rio-tiler/pull/782)
+
+    ```python
+    from rio_tiler.models import ImageData
+
+    img = ImageData(numpy.zeros((3, 256, 256), crs=CRS.from_epsg(4326), dtype="uint8"))
+    img_3857 = img.reproject("epsg:3857")
+    ```
+
 * add `indexes` parameter for `XarrayReader` methods. As for Rasterio, the indexes values start at `1`.
 
     ```python

pyproject.toml

@@ -29,7 +29,7 @@ dependencies = [
     "morecantile>=5.0,<7.0",
     "pydantic~=2.0",
     "pystac>=0.5.4",
-    "rasterio>=1.3.0",
+    "rasterio>=1.4.0",
     "color-operations",
     "typing-extensions",
     "importlib_resources>=1.1.0; python_version < '3.9'",

rio_tiler/models.py

@@ -14,13 +14,13 @@
 from rasterio.coords import BoundingBox
 from rasterio.crs import CRS
 from rasterio.dtypes import dtype_ranges
-from rasterio.enums import ColorInterp
+from rasterio.enums import ColorInterp, Resampling
 from rasterio.errors import NotGeoreferencedWarning
 from rasterio.features import rasterize
 from rasterio.io import MemoryFile
 from rasterio.plot import reshape_as_image
-from rasterio.transform import from_bounds
-from rasterio.warp import transform_geom
+from rasterio.transform import array_bounds, from_bounds
+from rasterio.warp import calculate_default_transform, reproject, transform_geom
 from typing_extensions import Self
 
 from rio_tiler.colormap import apply_cmap
@@ -34,6 +34,7 @@
     IntervalTuple,
     NumType,
     RIOResampling,
+    WarpResampling,
 )
 from rio_tiler.utils import (
     _validate_shape_input,
@@ -786,3 +787,44 @@ def get_coverage_array(
         ).astype("float32")
 
         return cover_array.sum(-1).sum(1) / (cover_scale**2)
+
+    def reproject(
+        self,
+        dst_crs: CRS,
+        resolution: Optional[Tuple[float, float]] = None,
+        reproject_method: WarpResampling = "nearest",
+    ) -> "ImageData":
+        """Reproject data and mask."""
+        dst_transform, w, h = calculate_default_transform(
+            self.crs,
+            dst_crs,
+            self.width,
+            self.height,
+            *self.bounds,
+            resolution=resolution,
+        )
+
+        destination = numpy.ma.masked_array(
+            numpy.zeros((self.count, h, w), dtype=self.array.dtype),
+        )
+        destination, _ = reproject(
+            self.array,
+            destination,
+            src_transform=self.transform,
+            src_crs=self.crs,
+            dst_transform=dst_transform,
+            dst_crs=dst_crs,
+            resampling=Resampling[reproject_method],
+        )
+
+        bounds = array_bounds(h, w, dst_transform)
+
+        return ImageData(
+            destination,
+            assets=self.assets,
+            crs=dst_crs,
+            bounds=bounds,
+            band_names=self.band_names,
+            metadata=self.metadata,
+            dataset_statistics=self.dataset_statistics,
+        )

tests/test_models.py

@@ -466,3 +466,71 @@ def test_image_encoding_error():
     """Test ImageData error when using bad data array shape."""
     with pytest.raises(InvalidFormat):
         ImageData(numpy.zeros((5, 256, 256), dtype="uint8")).render(img_format="PNG")
+
+
+def test_image_reproject():
+    """Test basic reproject functionality."""
+    data = numpy.zeros((1, 256, 256), dtype="uint8")
+    data[0:256, 0:256] = 1
+    mask = numpy.zeros((1, 256, 256), dtype="bool")
+    mask[0:100, 0:100] = True
+
+    # Create test image with WGS84 CRS
+    src_crs = CRS.from_epsg(4326)
+    img = ImageData(
+        numpy.ma.MaskedArray(data=data, mask=mask),
+        crs=src_crs,
+        bounds=(-95, 43, -92, 45),
+        metadata={"test": "value"},
+        band_names=["band1"],
+    )
+
+    # Test re-projection to Web Mercator
+    dst_crs = CRS.from_epsg(3857)
+
+    reprojected = img.reproject(dst_crs)
+    assert reprojected.crs == dst_crs
+    assert reprojected.count == 1
+    assert reprojected.width != 256
+    assert reprojected.height != 256
+    assert reprojected.array[0, 0, 0].data == 0
+    assert reprojected.array.data[0, -10, -10] == 1
+    assert reprojected.array.mask.shape[0] == 1
+    assert reprojected.array.mask[0, 0, 0]
+    assert not reprojected.array.mask[0, -10, -10]
+    assert reprojected.metadata == img.metadata
+    assert reprojected.band_names == img.band_names
+
+    # Test no re-projection when CRS is the same
+    same_crs = img.reproject(src_crs)
+    assert same_crs.crs == src_crs
+    assert same_crs.transform == img.transform
+    numpy.testing.assert_array_equal(same_crs.array, img.array)
+
+    # Test with different resampling method
+    reprojected_bilinear = img.reproject(dst_crs, reproject_method="bilinear")
+    with numpy.testing.assert_raises(AssertionError):
+        numpy.testing.assert_array_equal(reprojected_bilinear.array, img.array)
+
+    # With MultiBands
+    data = numpy.zeros((3, 256, 256), dtype="uint8")
+    data[:, 0:256, 0:256] = 1
+    mask = numpy.zeros((3, 256, 256), dtype="bool")
+    mask[:, 0:100, 0:100] = True
+
+    img = ImageData(
+        numpy.ma.MaskedArray(data=data, mask=mask),
+        crs=src_crs,
+        bounds=(-95, 43, -92, 45),
+    )
+
+    reprojected = img.reproject(dst_crs)
+    assert reprojected.crs == dst_crs
+    assert reprojected.count == 3
+    assert reprojected.width != 256
+    assert reprojected.height != 256
+    assert reprojected.array.data[:, 0, 0].tolist() == [0, 0, 0]
+    assert reprojected.array.data[:, -10, -10].tolist() == [1, 1, 1]
+    assert reprojected.array.mask.shape[0] == 3
+    assert reprojected.array.mask[:, 0, 0].tolist() == [True, True, True]
+    assert reprojected.array.mask[:, -10, -10].tolist() == [False, False, False]