Lecture 5&6 Flashcards
Frequencies
plausible building blocks of images and sounds
Fourier analysis
analyse the frequency characteristics from a signal
Fourier synthesis
the construction of a periodic signal on the basis of Fourier coefficients which gave the amplitude and phase of each component sine wave
Fourier transform
decomposition of a signal into it’s elements, measures the amplitude and the phase of different frequencies
which are the basics building blocks
sine and cosine waves
what is represented by the spectral signal
the strengths (magnitudes) of the constituent frequencies
Invertical transformation
going from the image domain to the fourier domain and vice versa
magnitude
the strengths of individual frequencies
phase
positions of the individual frequencies
what represents the DC point
the average intensity of the image
Interpreting magnitude plots
low spatial frequencies: closer to the DC point
Pick the correct term
The phase/ The magnitude
is more important for retrieving the content of an image
the phase
why the phase is more important when retrieving an image
because the phase captures the contours of tha image
Low-pass filtering
- blocks out the DC points and the high spatial frequencies
- let all low spatial frequencies to pass
Result: blurred image (bc the high spatial frequencies emphasizes the details)
High-Pass filtering
-blocks out the DC point and the low spatial frequencies
- let all high spatial frequencies pass
Result: Contour Image
Band-pass filtering
allow a range of spatial frequencies and set everything else to 0
(blocks the low and high spatial frequencies)
Fast convolution
perform a fourier transform on the template of the image
Does the human visual system perform a Fourier Transform? Why?
No.
Because the human visual system has multiple channels sensitive to different spatial frequencies, The Fourier transform is a single channel model.
Wavelet method
Filters take localised measurements of spatial frequencies
Texton method
Take patches of image -> perform clustering on patches
