Trying to "manually" convert to hsv a bgr image in opencv-python

Question

The algorithm I'm using is:

def rgb2hsv(r, g, b):
    r, g, b = r/255.0, g/255.0, b/255.0
    mx = max(r, g, b)
    mn = min(r, g, b)
    df = mx-mn
    if mx == mn:
        h = 0
    elif mx == r:
        h = (60 * ((g-b)/df) + 360) % 360
    elif mx == g:
        h = (60 * ((b-r)/df) + 120) % 360
    elif mx == b:
        h = (60 * ((r-g)/df) + 240) % 360
    if mx == 0:
        s = 0
    else:
        s = df/mx
    v = mx
    return h, s, v



def my_bgr_to_hsv(bgr_image):
    height, width, c = bgr_image.shape

    hsv_image = np.zeros(shape = bgr_image.shape)

    #The R,G,B values are divided by 255 to change the range from 0..255 to 0..1:

    for h in range(height):
        for w in range(width):
            b,g,r = bgr_image[h,w]
            hsv_image[h,w] = rgb2hsv(r,g,b)      

    return hsv_image

The problem that I'm getting is that when I want to display the image, I get only a black screen.

This is how I'm trying to display the image:

cv.imshow("hello", cv.cvtColor(np.uint8(hsv_image), cv.COLOR_HSV2BGR))

As you can see I convert it back to bgr in order to use cv.imshow, as it only uses bgr.

I don't think I understand enough of opencv or numpy to debug it.

Simply using imshow, shows the original picture in the wrong colors, which makes me think it can't be completely wrong.

I'm about to go out, so no time to check, but I believe OpenCV stores Hue values halved, i.e. between 0 and 180 (not between 0 and 260) so that they fit into an unsigned byte, so that may be biting you when you convert back. — Mark Setchell
Your HSV outputs are floats in 0..359 range. You treat then as an array of uint8, that is, of bytes. I suppose that the result of such a treatment would be strange at best. If you want to store such an image on disk, consider using Pillow, or at least find out the image's desired format. — 9000

Mark Setchell Mark Setchell · Accepted Answer · 2018-05-30T21:18:03

Your Hue values are scaled on the range 0 to 359, which is not going to fit into an unsigned 8 bit number, and your Saturations and Values are scaled on the range 0 to 1 which doesn't match your Hues for scale and is going to result in everything rounding to zero (black) when you cinvert it to an unsigned 8 bit number.

I suggest you multiply Saturation and Value by 255 and divide your Hue by 2.

Trying to "manually" convert to hsv a bgr image in opencv-python

1 Answers