imaging + signalling Flashcards
formation of real image by thin converging lens
lens changes curvature of incident wavefront
curvature = 1 / r
what is power of converging lens?
power is curvature added to wavefronts as they pass through converging lens
p = 1 / f
lens equation:
1/v = 1/u + 1/f
(u is negative)
where focus is, based on object / source position:
1) light from object closer to lens than fl: focus beyond fp
2) light from distant object (parallel incoming rays): focus at fp
3) light from near object but beyond fl (at distance u): focus at distance v
4) light from object at fl: focus at long distance (rays are made parallel)
linear magnification
m = image height / object height
on magnification diagram, triangles formed by paths of light rays before and after lens are congruent
magnification values:
1) negative: image is inverted
2) m < 1: image is smaller + closer to lens than object (source)
3) m = 1: image is same size + distance from lens to object (source)
4) m > 1 = image is larger + further from lens than object (source)
wave speed equation:
v = f x λ
frequency equation:
f = 1 / T
T: period of wave
characteristics of em waves (polarising):
1) electromagnetic waves are transverse + when unpolarised they oscillate in randomly changing plane
2) when polarised they oscillate in 1 fixed plane
how to polarise microwaves:
1) using metal grate + microwave transmitter (+ detector), if wave oscillates in plane parallel to grate alignment, wave is absorbed as electrons in grate can move length of grate so can absorb higher energy photons
2) if grate is rotated 90° so wave oscillates perpendicular to grate alignment, wave passes through as electrons in metal grate can only move width of each bar of grate - so, it cannot absorb photons of microwave as their energy is too large
pixel def:
pixel: single ‘picture element’ created by light sensitive detectors
bit def:
bit: smallest unit of digital information
byte
1 byte = 8 bits = 256 alternatives
resolution (imaging + signalling):
resolution: scale of smallest detail that can be distinguished
r = width of object in image / number of pixels across it
image processing:
1) changing brightness: add (+ / -) value on each pixel (to inc brightness, inc value on each pixel until brightest is coded at 255)
2) smoothing (removing noise): each pixel is replaced with mean of surrounding pixels
3) noise reduction (removing noise): each pixel is replaced with median of surrounding pixels
4) edge detection: mean of surrounding 8 pixels is subtracted from each pixel
5) changing contrast: to change contrast, multiply each pixel by (+ / -) value (where|value| < 1). to improve contrast, stretch range of pixels in image to full range (256) (values tend to 0 or 255 depending on which they are closer to) (image with little contrast does not use full range of pixels)
signal
signal transfers information from one place to another
(can be coded into binary digits to form digital signal)
analogue signal + cons:
analogue signal: continuously varying signal
1) can be amplified when distorted by noise, however this amplifies noise too
2) noise can be filtered out, but this loses signal clarity
digital signal + pros + cons:
digital signal: signal that only contains two values / ‘modes’, 0 or 1 / on or off
pros:
1) signal doesn’t lose detail when noise is filtered out
2) easy to detect, only takes values of 0 or 1 so can be regenerated perfectly
3) travels faster
4) carries more information
cons:
1) digital signals / numbers can be changed or scrambled, threats to online banking
2) digitally enhanced images on films - achieve level of reality we can’t achieve
amount of information in image? [equation]
info in image amount = num of pixels x bits per pixel
equations for quantisation levels + num of bits per sample needed to store them?
N = 2^b
b = log2[N]
where:
N = number of quantisation levels
b = bits per sample
def for noise
noise is random variation on a signal
what is sampling?
sampling is the process where the displacement of a continuous [analogue] signal is measured at small Δt and turned into a string of binary numbers (samples)
what is sampling rate / sampling frequency?
sampling rate / frequency is the number of samples per second
how to sample varying signal accurately?
• to sample a varying signal accurately, the time between samples must be shorter than the time between when important changes in the signal occur (eg troughs, peaks)
• with a larger Δt, details of the original signal is lost => can cause aliasing
quantisation levels description:
quantisation levels description:
• the signal being quantised can be represented graphically, with voltage (amplitude) on the y-axis and time on the x-axis (levels are kind of like the notches on the y-axis)
• number of levels = 2^b
what happens when you increase the number of quantisation levels?
increasing the number of quantisation levels causes a better signal match, however it also increases the demand on storage and transmission
what happens if the spacing between subsequent [/ consecutive] quantisation levels is smaller than the size of the noise variation?
if the spacing between subsequent [/ consecutive] quantisation levels is smaller than the size of the noise variation, then digitising the signal is useless as it would detail the noise [/ sample the noise to a higher detail], not reduce it
equation for max number of quantisation / alternative levels?
max number of levels (2^b) = V total / V noise
equation for max number of bits per sample?
max number of bits per sample (b) = log2
what to do if b (number of bits per sample) is decimal?
• if b is a decimal, round up!
• this ensures there are enough bits to provide the number of levels required
what is a quantisation error?
a quantisation error is the difference between the signal value and the quantisation level value
what is resolution (in signalling context)?
resolution is the smallest change in pd that can be determined
what is the equation for resolution (in signalling context)?
resolution =
pd range of signal / number of quantisation levels
what are the two conditions for sampling?
the two conditions for sampling are:
• the signal cannot contain frequencies above a certain maximum
• the min sampling rate must be at least 2x the max frequency of the signal (the Nyquist rate / Nyquist’s rule)
what is the standard sampling frequency for music?
the standard sampling frequency for music is 44.1kHz
(the limit of human hearing is 20 kHz, so the sampling frequency must be greater than 40 kHz)
what happens when frequencies above a certain maximum are sampled?
if frequencies above a certain maximum are sampled, the signal won’t be regenerated accurately and aliases will be generated
what happens if the sampling frequency used to sample a signal is too low [/ slow]?
if sampling is too slow, high frequencies will be missed and false, low frequency aliases will be created
what are aliases?
aliases are low frequency values that were not in the original signal
what is bit rate?
bit rate is the rate of transmission of digital information (units: Hz or bits^-1)
what is the question for bit rate?
the equation for bit rate is:
bit rate = samples per second x bits per sample
Note:
samples per second = sampling rate or sampling frequency
what is the equation for signal duration?
the equation for signal duration is:
signal duration =
number of bits per signal / bit rate