Image noise

Question 1

What are examples of uncorrelated noise?

Answer 1

• Shot noise & thermal noise in sensor (produces speckled pattern)
• Silver halide grains in film photography
• Neuronal noise in retina
• Quantization noise in digital photographs

Question 2

What does increasing exposure do?

Answer 2

Increasing the exposure to a small number of photons will decrease the number of speckles.

Question 3

What causes correlated noise? (interference)

Answer 3

• Electrical interference.
• Source/sensor interference.

Question 4

How can correlated noise be removed?

Answer 4

Using frequency domain

Question 5

What does correlated noise create?

Answer 5

Halftone distortion/moire patterns.

Question 6

Why is noise most commonly removed?

Answer 6

Aesthetic reasons

Question 7

How can noise be helpful?

Answer 7

Noise can be helpful in photo forensics and can act as a ‘photo fingerprint’ e.g linking a photo to a certain camera.

Question 8

What pattern does shot and thermal noise tend to follow?

Answer 8

Gaussian

Question 9

How is uniform noise created?

Continuous range

Answer 9

• Uniform noise is introduced when you quantize the intensity values.
• The camera isn’t recording a continuous range of intensity values from black to white so during the process of blocking up the intensities uniform noise is created.

Question 10

How can you remove Gaussian noise?

Answer 10

• Smooth the image, however, this can soften the edges of the image. The quality can also be worse than when the filter was applied.
• An alternative filter is the non-local means.

Question 11

How can you remove salt and pepper noise?

Answer 11

Using a median filter

Question 12

What does each camera have that’s unique?

Answer 12

• A noise pattern
• If you extract a pattern from a camera and a photograph and they have a high correlation then there’s a high chance that the photo was taken with that camera.