Resizing image without distortion creates a square background but in black

Question

In order to maintain the aspect ratio of my image I'm making use of the following code to create a square block and applying my image over this which is fine but the problem is the leftover background is dark/black is there a way to keep that white/transparent?

my code:

def resize_image(img, size=(28,28)):

h, w = img.shape[:2]
c = img.shape[2] if len(img.shape)>2 else 1

if h == w: 
    return cv2.resize(img, size, cv2.INTER_AREA)

dif = h if h > w else w

interpolation = cv2.INTER_AREA if dif > (size[0]+size[1])//2 else 
                cv2.INTER_CUBIC

x_pos = (dif - w)//2
y_pos = (dif - h)//2

if len(img.shape) == 2:
    mask = np.zeros((dif, dif), dtype=img.dtype)
    mask[y_pos:y_pos+h, x_pos:x_pos+w] = img[:h, :w]
else:
    mask = np.zeros((dif, dif, c), dtype=img.dtype)
    mask[y_pos:y_pos+h, x_pos:x_pos+w, :] = img[:h, :w, :]

return cv2.resize(mask, size, interpolation)

Answer 1

The background color is the color you initialize it with. (mask = np.zeros...) To have a white background use np.ones and for a specific color add a line after initialization mask[:, :] = (255, 0, 0).

Answer 2

You need to modify the mask you are using. It is initialized with zeros making the background black.

In the following snippet I have inverted the mask using cv2.bitwise_not() and used the result as background.

def resize_image1(img, size=(28,28)):
    h, w = img.shape[:2]
    c = img.shape[2] if len(img.shape)>2 else 1
    if h == w: 
        return cv2.resize(img, size, cv2.INTER_AREA)
    dif = h if h > w else w
    interpolation = cv2.INTER_AREA if dif > (size[0]+size[1])//2 else cv2.INTER_CUBIC
    x_pos = (dif - w)//2
    y_pos = (dif - h)//2
    if len(img.shape) == 2:
        mask = np.ones((dif, dif), dtype=img.dtype)
        mask = cv2.bitwise_not(mask)               # Added mask inversion here
        mask[y_pos:y_pos+h, x_pos:x_pos+w] = img[:h, :w]
    else:
        mask = np.ones((dif, dif, c), dtype=img.dtype)
        mask = cv2.bitwise_not(mask)               # Added mask inversion here
        mask[y_pos:y_pos+h, x_pos:x_pos+w, :] = img[:h, :w, :]
    return cv2.resize(mask, size, interpolation)

Sample output of image resized to 150 x 150 dimensions: