Review: Pixels and Filters

You've covered a huge amount, so before the capstone, let's make it stick. Research on learning is clear. Testing yourself is far more effective than rereading. So these review lessons mix a quick recap with questions. Don't skip them. Pause on every question, and commit to an answer.

• Testing beats rereading
• A recap, then questions
• Pause on every one

Recap: images and filters

An image is a grid of numbers. Colour spaces such as HSV separate colour from brightness, which makes thresholds far more reliable. A filter slides a small kernel across the image. Blur averages, and edge filters measure change. Thresholds turn images into masks, morphology cleans those masks up, and contours measure the shapes.

Question 1

Question 1. Why use HSV for picking out red apples? A, It is smaller. B, It separates colour from brightness. C, It removes noise. D, It is faster. Pause and choose. The answer is B. Lighting changes affect colour far less there.

Question 2

Question 2. What's a kernel? A, A small grid that slides over the image. B, A kind of camera. C, A file format. D, A neural network. Pause and choose. The answer is A. Each position gives one output value.

Question 3

Question 3. Which removes small specks from a mask? A, Dilation alone. B, Opening. C, Equalising. D, Resizing. Pause and choose. The answer is B. Opening erodes, then dilates.

Question 4

Question 4. Why is red tricky in HSV? A, It has no saturation. B, It wraps around the colour wheel. C, It is too bright. D, It has no hue. Pause and choose. The answer is B. You need two ranges.

Question 5

Question 5. Which fixes uneven lighting best? A, Global equalisation. B, CLAHE. C, Sharpening. D, JPEG. Pause and choose. The answer is B. It works in small tiles, with a limit.

Question 6

Question 6. What does the variance of the Laplacian measure? A, Colour. B, Sharpness. C, Size. D, Noise only. Pause and choose. The answer is B. Blurry images have soft edges.

Summary