A quiz on the first laboratory: Getting started with OpenCV

Adrian F. Clark

1. What is the standard deviation?

   • The same as the mean.
   • How different one image is from another.
   • The median.

   • The average spread of values from the mean.

   The standard deviation measures the average spread of value in an image.

2. What do the three subscripts of a colour image represent in OpenCV?

   • colour, row, column.
   • column, row, colour.
   • colour, column, row.

   • row, column, colour.

   Images are stored as numpy matrices, so their first two subscripts are row and column respectively. Images are stored so that the colour values of a single pixel are next to each other in memory, so the third subscript selects the colour.

3. What values did your extrema routine return for testimage?

   • 0, 10.

   • 10, 15.

   • 0, 15.
   • 0, 255.

   Your extrema routine should return the smallest and largest values in the image it is given. For testimage, these are 10 and 15 --- in fact, the doctest for this routine gives them.

4. What is the purpose of the doctest module?

   • To provide an alternative way of commenting code.

   • To let you store unit tests in the top-of-routine documentation.

   • To document a method.
   • To provide an overview of a complete module.

   The doctest module provides a way for a developer or tester to store a series of tests. These can be pulled out of the source code and executed automatically, checking that the routine produces when run exactly the documented output.

5. Why were you asked to store the routines you develop in sxcv.py?

   • To give you practice at programming.
   • So that you can see how to comment programs.
   • So you can record what you have done.

   • So that you can use them later in vision application programs.

   The underlying reason for getting you to store the routines you write (and hopefully test) in sxcv.py is so you can use them to extend the programs that appear in the lab experiments after the first progress test.

6. How are images represented in OpenCV?

   • as a list of unsigned bytes

   • as numpy arrays

   • as a list of lists
   • in an opaque OpenCV-only format

   The Python wrappers for OpenCV represent images as numpy arrays, 2D for monochrome ("black and white") images and 3D for colour ones.

7. What does cv2.imread do if it cannot read an image?

   • It outputs an error message and exits.
   • It crashes.
   • It throws an exception.

   • It returns None.

   OpenCV, or at least its Python wrapper, returns None when it cannot read an image. Good programs will catch this case and do something sensible...but not all programs are good.

8. What underlying OpenCV routine does sxcv.display call to display an image?

   • cv2.imshow

   • cv2.display
   • cv2.imread
   • cv2.waitKey

   The underlying display routine in OpenCV is cv2.imshow, though the image is not posted onto the screen until cv2.waitKey is called.

9. How are the pixels of an image normally represented in OpenCV?

   • unsigned ints

   • unsigned bytes

   • signed floats
   • signed bytes

   Unless you take care to make them otherwise, the pixels of images are stored as unsigned bytes in OpenCV. It is possible to use the other representations listed in the answers but you have to go to some effort to do so when you create the image.

10. Where is the origin of an image in OpenCV?

    • the bottom right corner
    • the bottom left corner

    • the top left corner

    • the top right corner

    OpenCV images have the origin (row zero, column zero) at the top left corner of the image. This might seem obvious but is not always the case: for example if you program image operations using the 3D graphics package called OpenGL (note the similarity of the name to OpenCV), its image origin is at the bottom left corner of an image. Much confusion can result from this!