I am trying to draw polygons on the microscopic images to highlight the important areas. I will skip the part how the coordinates of the vertices are obtained, because this is not important here. To draw polygons on the image I use a generated list of dictionaries with information about coronal section index cortical, layer number (1, 2, 3, 4, 5, 6), cortical area id value, and coordinates for given layer (so a set of vertices that correspond to particular layer with shape (Y, SLICE, X)). Here is how the list of dictionaries looks like: example_image The problem is, when I am trying to run the function responsible for drawing polygons, the process is taking very much time (above 2 hours). I understand that this is due to large size (~20000x30000 pixels), but is there any way to make this faster using skimage? I can't use OpenCV or Pillow. layer_colors = { 1: (255, 0, 0), # Red 2: (0, 255, 0), # Green 3: (0, 0, 255), # Blue 4: (255, 255, 0), # Yellow 5: (0, 255, 255), # Cyan 6: (255, 0, 255) # Magenta } def draw_full_res(image, coords, image_slice_idx, thickness=5, layer_colors=layer_colors): disk_kernel = ski.morphology.disk(thickness) mask = np.zeros(image.shape[:2], dtype=bool) for idx in range(len(coords)): if coords[idx]['slice'] == image_slice_idx: layer = coords[_idx]['layer'] color = layer_colors.get(layer) print('Progress layer: ', layer, '|', color) for contours in coords[idx]['coords']: # Draw only those polygons where number of vertices is greater than 5 if len(contours) > 5: # Because shape for coords is (Y, SLICE, X) y_coords = contours[:, 0].astype(int) x_coords = contours[:, 2].astype(int) # Create polygon rr, cc = ski.draw.polygon_perimeter(x_coords, y_coords, shape=image.shape[:2]) mask[rr, cc] = True # Make polygon thicker mask = ski.morphology.dilation(mask, disk_kernel) image[mask] = color output_path = f'{image_slice_idx}_slice.png' ski.io.imsave(output_path, image) Continue reading...
