bjnortier/crop-satellite-tiles-to-geojson

Repo for my blogpost on cropping tiles satellite imagery to a GeoJSON boundary

crop-satellite-tiles-to-geojson's Issues

Solve RGB in one step

crop-satellite-tiles-to-geojson/crop.py

Line 143 in 2eb6754

# How to do RGB in one step?

Use .polygon(fill=255) to draw the alpha mask with a 255 value
and later just merge the single sliced layer in masked_raw[...,3] = mask_raw

def mask_pixels_outside_boundary(cropped_to_boundary_image,
                                 boundary_coordinates,
                                 cropped_bbox):
    """
    Mask the image to only include pixels inside the GeoJSON boundary.
    """
    polygon_pixels = create_boundary_pixels(
        boundary_coordinates,
        cropped_bbox,
        cropped_to_boundary_image.size)
    mask_image = Image.new('L', cropped_to_boundary_image.size, 0)
    ImageDraw.Draw(mask_image).polygon(polygon_pixels, outline=0, fill=255)
    mask_raw = np.array(mask_image)

    cropped_to_bbox_raw = np.asarray(cropped_to_boundary_image)
    masked_raw = np.empty(cropped_to_bbox_raw.shape, dtype='uint8')
    masked_raw[...,3] = mask_raw
    masked_image = Image.fromarray(masked_raw, 'RGBA')
    return masked_image