I’ve been trying to pansharpen some multispectral images using Python within a Docker environment, but the output image quality is very poor. The pansharpened image appears degraded, and I’m not sure what I might be doing wrong. I would appreciate any guidance or suggestions.

1- Locate the Panchromatic Image: I first search for the panchromatic file in the directory using the following function:

def find_panchro_images(directory):
    panchro_images = []
    other_images = []

    for root, _, files in os.walk(directory):
        for file in sorted(files):  # Sort the files before processing
            if file.lower().endswith(".tif"):
                file_path = os.path.join(root, file)

                try:
                    result = subprocess.run(
                        ["exiftool", "-BandName", file_path], stdout=subprocess.PIPE, stderr=subprocess.PIPE, text=True
                    )

                    if "Panchro" in result.stdout:
                        panchro_images.append(file_path)
                    else:
                        other_images.append(file_path)
                except Exception as e:
                    print(f"Error processing {file_path}: {e}")

    return panchro_images, other_images

2- Stack the Multispectral Channels Together: After identifying the multispectral bands, I stack them using this function:

def read_and_stack_bands(files):
    bands = []
    for file in files:
        with rasterio.open(file) as src:
            band = src.read(1)
            bands.append(band)

    stacked_image = np.dstack(bands)
    return stacked_image

3- Pansharpening: I then use the pansharpen function from the GitHub repository https://github.com/ThomasWangWeiHong/Simple-Pansharpening-Algorithms for the pansharpening process:

def pansharpen(m_files, pan, psh, R=3, G=2, B=1, NIR=4, method="esri", W=0.05):
    # Function implementation...

4- Finally, I execute the following script to apply the pansharpening:

if __name__ == "__main__":
    for folder in ["photo_1", "photo_2"]:
        folder_path = os.path.join(yasmina_dir, folder)
        output_folder = os.path.join(yasmina_dir, "output", folder)
        os.makedirs(output_folder, exist_ok=True)

        panchro_images, other_images = find_panchro_images(folder_path)

        print(f"nIn folder '{folder}':")
        if panchro_images:
            print("Panchromatic Image:")
            for image in panchro_images:
                print(image)

            print("Other Images:")
            for image in other_images:
                print(image)

            if len(panchro_images) == 1:
                other_images = sorted(other_images)[:4]  # Use only [B, G, R, NIR]

                output_image_name = os.path.basename(other_images[0]).replace("_1.tif", "_pansharpened.tif")
                psh_output_path = os.path.join(output_folder, output_image_name)

                pansharpen(other_images, panchro_images[0], psh_output_path)
            else:
                print("Multiple Panchromatic images found. Skipping pansharpening.")
        else:
            print("No Panchromatic images found. Skipping pansharpening.")

When I check the output image, it looks very bad (e.g., overly pixelated, image in grayscale). I’m not sure if the issue is due to the pansharpening method I’m using, a problem with how I’m reading the images, or something else.

Has anyone encountered a similar issue or can suggest what might be going wrong? I’m particularly interested in understanding if there’s a flaw in my image processing pipeline or if the problem lies with how the images are being rescaled or processed before pansharpening.