cv2 edited image saves with wrong colors

Question

I'm working on a grayscale .tif file:

I convert it to BGR and try to draw some colorful stuff on it. If I save the result as .png, it's all still in shades of gray, including the drawn elements. If I save it as .jpg, colors of them are okay, but the rest of image is a lot brighter than it was, which I definitely don't want to happen.

simplified example of what I'm trying to do:

def draw_lines(image_path):
    image = cv2.cvtColor(cv2.imdecode(np.fromfile(image_path, dtype=np.uint8), cv2.IMREAD_UNCHANGED), cv2.COLOR_GRAY2BGR)

    cv2.line(image, (0,10), (image.shape[1], 1000), (0, 255, 0), 10)

    cv2.imwrite("result.jpg", image)

You can read the image simply by `cv2.imread` and then draw the line using `cv2.line` directly without converting colors and all. — krxat, Jul 30 '20 at 12:20
I need this to work with file paths containing unicode characters, and `cv2.imread` doesn't work properly with them — Pen Guin, Jul 30 '20 at 12:45
Try to use the function `cv2.imshow` to see how the image looks before saving. Is it in the correct format you wanted? — krxat, Jul 30 '20 at 12:46
sound like wrong pixel format (8 bit, 16 bit, ...or integer vs. float). i dont know how to test it in python. — Micka, Jul 30 '20 at 16:34
Please share your TIFF file. You might use Dropbox or Googe Drive for example. — Mark Setchell, Jul 30 '20 at 22:56

score 0 · Answer 1 · answered Jul 31 '20 at 01:55

I'm not exactly sure why, but apparently the problem was caused by usage of flag cv2.IMREAD_UNCHANGED After changing this line to image = cv2.imdecode(np.fromfile(image_path, dtype=np.uint8), cv2.IMREAD_COLOR) everything works fine, no matter which extension

cv2 edited image saves with wrong colors

1 Answers1