Using GDAL or Python, how can I merge/combine 3 NITF images (RGB) into one image and keep the RPC metadata intact?

Question

I'm looking to merge 3 imagery bands that are saved as separate files. The file format is NITF. The reason they are saved as NITF is because they have RPC metadata embedded into the images. I would like to combine the NITF images so the output image retains the RPC metadata. I will be doing this a lot so I would like to build a script either through Python or GDAL so that I don't have to do this manually. Unfortunately I'm very much a novice when it comes to scripting, and after scouring the internet I have hit a road block. I was able to successfully merge the images in the desired output using GDAL, but the RPC metadata did not transfer into the output image.

Here's the code I have so far:

gdal_merge.py -o output.ntf -of NITF -seperate red_band.ntf green_band.ntf blue_band.ntf -co PHOTOMETRIC=RGB

Alessio Arena · Answer

This should do the trick in Python. If you don't have rasterio install I suggest to get it through conda.

import numpy as np
import rasterio

data = []
input_images = [path1, path2, path3]
for img in input_images:
    # Open the image file, then read the data in memory as np.ndarray
    img_meta = rasterio.open(img)
    img_data = img_meta.read()
    data.append(img_data)

# stack along the first axis all data
new_data = np.vstack(tuple(data))

# This assumes that all of them have the same crs, transform and nodata value
newprofile = {
    'count': new_data.shape[0],
    'crs': img.crs,
    'dtype': 'uint8',
    'driver': 'GTiff',  # You can pick other formats that suits you. There is read-write support for NITF if you prefer
    'transform': img.transform,
    'height': new_data.shape[1],
    'width': new_data.shape[2],
    'blockxsize': 256,
    'tiled': True,
    'blockysize': 256,
    'nodata': img.nodata 
}

# Initialise a new GeoTiff and write the data
with rasterio.open(outpath, 'w', **newprofile) as dst:
    dst.write(new_data)

Regarding metadata I'm not sure how you want to combine them, but you can read them using img.tags(). This will return a dictionary, that you can write to the new image using the dst.update_tags(). Please note that this last function is only available if you open the image in 'w' or 'r+' mode

Using GDAL or Python, how can I merge/combine 3 NITF images (RGB) into one image and keep the RPC metadata intact?

One Answer

Add your own answers!

Ask a Question