GDAL 的 GetProjection() 和 GetGeoTransform() 的 Rasterio 等价物？

Question

I have a GeoTIFF that I've done a lot of processing with in OpenCV. After processing, its lost its spatial information so I need to insert it back in. In GDAL, I would just extract that from the original image before processing and then just shove it back in with the post-processed image like so:

#open original geotiff
original_img = gdal.Open('/path-to-file/original_img.tif', 1)
crs = original_img.GetProjection()
gt = original_img.GetGeoTransform()
del original_ds

#open cv processes here....write post-processed file

# open processed raster and set the projection and geotransform
img = gdal.Open('/path-to-file/processed_img.tif', 1)
img.SetProjection(crs)
img.SetGeoTransform(gt)

Is there a way I can do this in Rasterio?

Answer 1

According to RasterIO documentation , you can simply access crs and transform attributes on an open dataset to get the projection and geotransform respectively. You can then pass them back to output image file on opening .

Example:

# Open original Geotiff
original_img = rasterio.open("path/to/input.tif")
# Extract spatial metadata
input_crs = original_img.crs
input_gt  = original.transform

# Do your processing. For this example, just read first band of input dataset
processed_img = original_img.read(1)

# Prepare output geotiff file. We give crs and gt read from input as spatial metadata
with rasterio.open(
  'path/to/output.tif',
  'w',
  driver = 'GTiff',
  count = 1,
  height = processed_img.shape[0],
  width  = processed_img.shape[1],
  dtype  = processed_img.dtype,
  crs    = input_crs,
  transform = input_gt  
) as output:
  output.write(processed_img, 1)

Notes on above example :

• assume that image dimensions are preserved when processing it. If not, you'll have to derive the spatial transform accordingly.
• Assume single-band dataset