Ch. 1 - Introduction to Digital Image Processing Flashcards
1
Q
LSF
A
- Line Spread Function
- Describes system performance in one direction
- Can be approximated by the Gaussian Function
LSF(y) = [1/(2pi(sigma^2))^1/2]exp[-((y-y0)^2)/(2(sigma^2))]
2
Q
PSF
A
- Point Spread Function
- Can be used to measure the spatial resolution in 3D
3
Q
FWHM
A
- Full Width at Half Maximum
- Features are separated by a distance greater that the FWHM of the SF, then they can be resolved as separate structures as opposed to one larger structure
FWHM = (2((2ln2)^1/2))sigma =~ 2.36*sigma
4
Q
Spatial Resolution
A
- The smallest distance between two features such that the features can be individually resolved rather than as one larger shape
- Line Spread Function (LSF)
- Full Width at Half Maximum (FWHM)
- Point Spread Function (PSF)
- Modulation Transfer Function (MTF)
5
Q
SNR
A
- Signal - to - Noise Ratio
- Noise - measured as the standard deviation
- SNR = signal/noise
- True noise = random
- You want a high S/N –> a higher signal is better
- Can improve by averaging
6
Q
CNR
A
- Contrast - to - Noise Ratio
- Even if you have a good signal to noise ratio, it does not matter if the contrast to noise is too low
CNR(AB) = C(AB)/sigma(N) = |S(A)-S(B)|/sigma(N) =
= |SNR(A) - SNR(B)|
7
Q
Image Bit Depth
A
- 2-Bit ADC
- Another bit = twice as many digital levels
- Better resolution
- Another bit = twice as many digital levels
8
Q
Image Bit Depth
A
- 2-Bit ADC
- Another bit = twice as many digital levels
- Better resolution
- Another bit = twice as many digital levels
# of Levels = 2^x
- x = number of bits
Resolution =[V(in)max - V(in)min]/2^x
Gray-Scale Range
–> Range = Max-Min
9
Q
Filtering - Image Convolution
A
- Median Filter
- Mean Filter
- High Pass Filter
- Low Pass Filter
- Sharpening Filter
10
Q
1D Convolution
A
- An integral that expresses the amount of overlap in a filter “h” as it is shifted over function “f”
- ->Blends one function with another function(f*g)(t)=def=S(^infinity)(-infinity)f(tau)g(t-tau)d(tau)
-LOOK AT HOMEWORK
11
Q
Image Convolution
A
- Steps for convolution
- Mirror h about the axis and shift it by t
- Multiply x and h
- Integral (i.e. find area under the over lap)
- Repeat
12
Q
Image Characteristics - Image Quality
A
- Spatial Resolution
- Signal to Noise Ratio
- Image Calibration
- Contrast
- Quantitation
- Image Processing
13
Q
Modulation Transfer Function (MTF)
A
- Given by the Fourier Transform of the PSF
MTF(kx,ky,kz) = F{PSF(x,y,z)}
–> where kx,ky and kz are the spatial frequencies measured in lines/mm corresponding to the x,y, and z spatial dimensions measured in mm.
-The MTF is the commonly used measure of the spatial resolution of an imaging system
–> Called the Optical Transfer-Function (in the book)
14
Q
Image Calibration
A
Dark current subtraction
15
Q
Image Contrast
A
C(AB) = |S(A) - S(B)|
- -> S(A) & S(B) = signals from tissue A and B - -> C(AB) = contrast between tissue A and B
