Human Image Perception

Question 1

Light is focussed on the ?

Answer 1

Retina

Question 2

What are cone cells?

Answer 2

• located near the centre of the retina (fovea)
• Sensitive to colour and luminance
• Convert wavelengths of life into 3 values known as a tri-stimulus. (S (short), M (middle), L (long) cone cells)

Question 3

What are rods?

Answer 3

• Located near the periphery of the retina
• Much more sensitive to light, luminance only, more sensitive to motion, less resolution

Question 4

What is the visual cortex?

Answer 4

• Filtering, object recognition, edge detection, stereo fusion
• Electrical Impulses from the retina are channelled by the optic nerve to the Visual Cortex

Question 5

What is metamerism?

Answer 5

Light is a combination of wavelengths. Different combinations may appear to be the same colour. (two colours appear to match under one lighting condition, but not when the light changes)

Question 6

CIE RGB values are…

Answer 6

red = 700 nm
green = 546.1 nm
blue = 435.8 nm

Question 7

What are colour matching functions?

Answer 7

Colour matching functions - r̄(λ), ḡ(λ) and b̄(λ) - represent the amounts of light from the red, green and blue primaries, in tristimulus units, needed to match unit intensity of light with a narrow band of wavelengths centred on λ. Can be used to estimate RGB values for any combination of colours. The functions were obtained by allowing participants to combine the 3 r,g,b stimuli to match the appearance of colour from a single wavelength

Question 8

CIE XYZ means…

Answer 8

Y = roughly similar to response of Cones M (green stimulation)
Z ≈ Cones S (blue stimulation)
X is a mix of cone responses
- linear transformation of RGB
- standard reference for building other colour spaces (no negative values)

Question 9

What is the human gamut?

Answer 9

All colours visible by an average human on the x-y plane. The chrominaticity diagram is used for comparing the gamuts of different colour spaces.

Question 10

What is the spectral locus?

Answer 10

In chromacity diagram, all monochromatic lights lie on spectral locus

Question 11

HSV colour space means…

Answer 11

H = hue (shade of colour)
S = saturation (colour depth)
V = visual (brightness of colour)

Question 12

YUV colour space means…

Answer 12

Y = luminescence
U,V = chrominance

Question 13

Spatial frequency

Answer 13

N/θ
tan(θ) = (N x cycle period)/(viewing distance)

Question 14

Weber’s Law for threshold visibility

Answer 14

k = ∆I/I
Weber’s law relates the perceived brightness of an object to the brightness of its background. Implies that you need more Brightness difference to resolve an object against a bright background than against a dark background.

Question 15

Mach banding

Answer 15

The HVS perceives that each vertical stripe looks brighter on the left and darker on the right. It is as a direct result of spatial filtering in the visual cortex

Question 16

Activity masking

Answer 16

• Image processing method in which we define a small ‘image piece’ and use it to modify a larger image. Masking is the process that is underneath many types of image processing, including edge detection, motion detection, and noise reduction.

Question 17

Statistical redundancy

Answer 17

In local image regions, say 8x8 blocks, the data tends to be flat or typically homogenous much of the time. This redundancy can be removed without affecting the image substantially.

Question 18

HVS response

Answer 18

HVS response to image stimuli implies that one can introduce artefacts into images without them being seen. The colour subsampling illustrated this idea. Thus techniques that remove statistical redundancy can apply that concept heavily in regions where the resulting defects will not be noticed.

Question 19

Efficient coding techniques can be used to…

Answer 19

• Represent any data as a more compact stream of digits.
• Compression and error-resilience.

Question 20

Objective assessment of image quality

Answer 20

• Mean Squared Error (MSE)
• Mean Absolute Error (MAE)
• Signal-to-noise ratio (SNR)
• Peak signal-to-noise ratio (PSNR)
• Structural Similarity Index Metric (SSIM):
• VMAF (video)