A quiz on the colour laboratory

Adrian F. Clark

1. What colour correspond to OpenCV HSV values (120, 0, 100)?

   • grey

   • red
   • blue
   • green

   The saturation is zero, so this is lies on the black-to-white axis. Its value is 100, so it is a mid-grey.

2. Why is segmenting red difficult in HSV colour space?

   • HSV represents only complementary colours, as used in CMYK
   • the hue is about 180 degrees
   • the hue is negative

   • hues corresponding to red lie just above zero AND just below 360 degrees

   Red hues have values of zero up to about 30 degrees AND from about 330 to 360 degrees, meaning you need two separate masks with cv2.inRange to isolate them.

3. How would two masks returned from cv2.inRange be combined?

   • by XOR-ing them
   • by subtracting them

   • by OR-ing them

   • by AND-ing them

   The easiest way to combine masks from cv2.inRange is to OR them.

4. Given a three-channel colour image as input, what does cv2.inRange return?

   • a 2D image set to the grey values of the pixels inside the range
   • a single-colour image

   • a 2D image set to zero and 255

   • a 2D image set to the grey values of the pixels outside the range

   cv2.inRange returns a 2D image with pixels set to 255 where they lie in range and zero otherwise.

5. What ordering are the colour channels by default in OpenCV?

   • RBG
   • BRG

   • BGR

   • RGB

   For reasons best (or maybe only) known to its developers, OpenCV orders its channels as blue, green, red. Weird, isn't it?

6. What data type are the colour arguments in a call to cv2.inRange?

   • integer
   • a list
   • float

   • a tuple

   You normally write the colours as a triplet of integer values in round brackets, and that makes them into what Python calls a tuple.

7. What colour corresponds roughly to HSV values (60, 90, 90)?

   • cyan
   • purple

   • yellow

   • blue

   The hue is roughly mid-way between red and green, so the shade is yellow. The saturation and value are both high, so this is actually a bright yellow.

8. Why are the HSV values used by most people changed for use in OpenCV?

   • to avoid the use of percentages

   • to fit into an unsigned byte

   • to make everyone's life difficult
   • to execute more quickly

   RGB-to-HSV conversion is performed in-place by cv2.cvtColor, so the values need to be able to fit into the same unsigned byte representation as the original pixels.

9. Which OpenCV HSV values are closest to (180, 50, 50)?

   • (90, 128, 128)

   • (360, 100, 100)
   • (360, 255, 255)
   • (0, 0, 0)

   You can work this one out just by looking at the numbers: they are half-way through the possible ranges in each case, so the answer is the middle of the ranges used by OpenCV.

10. What does saturation mean in HSV colour space?

    • how bright the colour is compared to black

    • how intense the colour is compared to grey

    • what the colour is
    • how many bits are needed to represent it

    Saturation is how far the colour lies away from the grey line (the black-to-white axis of the colour cone), with colourse becoming more intense as the saturation moves from grey on the axis to fully-saturated at the edge of the cone.