Potential test case: normalise image to its intensity STD and AVG

[Romain-Laine]

There are many ways to normalise an image to 0-1 range. One fun way is to use the image intensity STD and AVG to do that, here's a snippet.

def std_avg_normalisation(img: np.ndarray) -> np.ndarray:
    """
    Arguments:
    - img: input image to normalise
    Output:
    - np.array: the normalised version of the image, as np.float32
    """

    # Convert the data to float32
    img = img.astype(np.float32)

    img_mean = np.mean(img)
    img_std = np.std(img)
    return (img - img_mean)/(4*img_std) + 0.5