Lecture 3/4 Display tech / Image formation and filtering Flashcards
Why is a barrier needed between an object and film in a pin-hole camera?
To control the rays of light from the object
Only allowing light to enter through a pin-hole aperture.
Dimensionality Reduction Machine (?? to ??)
What does this machine accomplish?
Convert the 3D world to a 2D image
How does a large aperture affect an image?
Blurry and bright image
How will having a tiny aperture affect an image?
Too small an aperture will create a blurry and dark image
What does a lens do to an image?
A lens focuses parallel rays of light onto film
- Objects focus at a distance of f
- Other points project to a “circle of confusion” in the image
- Changing the shape of the lens changes the focal distance
How does focal distance relate to object and image distance?
f: focal distance
u: object distance
v: image distance
What shape does a lens take?
Lenses are typically spherical - this is easier to produce
What is D?
The aperture of diameter D restricts the range of rays
- aperture may be on either side of the lens
What is f?
The focal point is at a distance f beyond the plane of the lens
- f is a function of the shape and index of refraction of the lens
What are the effects of changing the aperture size?
Changing the aperture size affects the depth of field
- A smaller aperture increases the range in which the object is approximately in focus
Compare a camera to the human eye
Iris - colored annulus with radial muscles
Pupil - the hole (aperture) whose size is controlled by the iris
What’s the “film” equivalent in an eye?
photoreceptor cells (rods and cones) in the retina
What does this image (from a sensor array) show?
A continuous image projected onto a sensor array
What does this image (from a sensor array) show?
Result of image sampling and quantisation
How does a sensor array record light?
▪ turn continuous light into pixels
▪ cells detect photons (as electrons)
over an array of pixels
What is a CCD?
Stands for Charge Coupled Device. A type of image sensor.
They are serial devices where pixels
are read out one at a time.
What is a CMOS?
A type of image sensor.
each pixel contains an
amplifier, so read-out can be faster (than a CCD)
Why are there different image sensor types, what are they?
Different sensor types use different
approaches to read out electrons.
CCD - pixels read out 1 by 1
CMOS - each pixel contains an amplifier… faster read-out
What are the active image display technologies?
LCD - Liquid Crystal Display
Digital Mirror Displays
OLED
Plasma Displays
Cathode Ray Tube
What are passive image display technologies?
PRINTERS:
Laser printers
ink-jet printers
Commercial printers
What active image display is most common today?
LCD - Liquid Crystal Display
What active image display was standard for most of the 20th century?
Cathode Ray Tube
What are OLEDs used for?
Expensive high end devices.
Emit RGB no need for a back light (like LCDs)
What is a negative of plasma displays?
They’re power-hungry.
They were briefly popular.
How do LCDs work?
liquid crystals twist the
polarisation of light
- voltage across the liquid crystal
controls how much light
passes:
– voltage on: no light
(crystals align the electric field)
– Voltage off: maximum light
(crystals align to each other,
smoothly twisting light)
How is light controlled when voltage is ON in an LCD display?
ON = No light
crystals align with the electric field
How is light controlled when voltage is OFF in an LCD display?
OFF = Maximum light
crystals align to each other,
smoothly twisting light
How does the effect of quantisation vary as the bit number is increased?
More continuous and detailed image with more bits
How many bits (colour depth) are
needed in a qunatised image?
What does error diffusion achieve
Error diffusion achieves visually more pleasing results by distributing
quantisation errors across neighbouring pixels
How would you increase the brightness of an image?
Increase the value of each pixel by a fixed amount such that all pixel values are equal to or less than 255 (max)
In per pixel processing what is the image inversion equation?
Inverted Image
In per pixel processing what is the contrast enhancing equation?
What are these point-based filter maps showing?
Darkening and lightning
What are these point-based filter maps showing?
High and low contrast
What are these point-based filter maps showing?
Compressed to darks and compressed to lights
What are these point-based filter maps showing?
Emphasise shadows and lights
What are point-based filters?
These filters operate on a per-pixel basis.
They involve applying a function to each pixel of an image individually, based on the value of that pixel.
Common point-based filters: linear, median, and convolutional filters.
What are the limitations of a point-based filter?
Limited ability to capture structure and context: causes degraded image quality or loss of important image features.
Sensitivity to noise: result in the amplification of noise and degraded image quality.
Limited control: they apply a fixed function to each pixel. This can make it difficult to fine-tune the processing.
Over- or under-processing: unnatural or distorted appearance.
How to overcome noise in point-based filters?
Use neighbourhood filters
What filter can be applied to an image to remove salt and pepper noise?
Median filter: this will replace each pixel with the median value of all pixels in
its neighborhood. NOT A CONVOLUTION FILTER
What is the difference between correlation and convolution?
Convolution is
correlation with the
kernel rotated 180°
In a morphological filter, we can obtain erosion of a pixel using the following formula
In a morphological filter, we can obtain the dilation of a pixel using the following formula
In a morphological filter describe Opening
DILATION AFTER EROSION
In a morphological filter describe Closing
EROSION AFTER DILATION
What are examples of sobel filters?
Horizontal gradients
Vertical gradients
Diagonal gradients
What does a sobel filter kernel look like for:
Horizontal gradients
What does a sobel filter kernel look like for:
Vertical gradients
What does a sobel filter kernel look like for:
Diagonal gradients
Show equation for a gaussian (low pass) filter
What are the fundamental differences between a box filter and a gaussian filter?
Box filter gives equal weight to all pixels
Gaussian trusts the center pixel more
Visualise a kernel for
A motion blur convolution filter
Visualise the convolution kernel:
Sharpening using unsharp masking