Lecture9 digital image processing for notes
Lecture9 digital image processing for notes
(a) (b)
Figure 9.1 Functions of (a) Fourier transform and (b) Wavelet transform
Image Pyramid
is a powerful simple structure for representing images at more than one
resolution. It is a collection of decreasing resolution images arranged in
the shape of a pyramid as shown in the figure below.
Subband Coding
is used to decompose an image into a set of bandlimited components
called subbands, which can be reassembled to reconstruct the original
image without error. Each subband is generated by bandpass filtering the
input image. The next figures show 1D and 2D subband coding.
For example, if f={f1,f2,f3,f4 ,f5 ,f6 ,f7 ,f8 } is a time-signal of length 8, then
the HWT decomposes f into an approximation subband containing the
Low frequencies and a detail subband containing the high frequencies:
Low = a = { + , + , + , + }/√2
High = d = { − , − , − , − }/√2
The next figure shows an image decomposed with 3-level Haar wavelet
transform.
(b) Level 1
(c) Level 2
(d) Level 3
Figure 9.6 Example of a Haar wavelet transformed image
(a) (b)
(c) (d)
Figure 9.7 Histogram of (a) LL-subband (b) HL-subband (c) LH-subband (d) HH-subband of
subbands in Figure 9.6 (b)
(a)
(b)
Figure 9.8 (a) gray image. (b) its one-level wavelet transform
Note the horizontal edges of the original image are present in the HL
subband of the upper-right quadrant of the Figure above. The vertical
edges of the image can be similarly identified in the LH subband of the
lower-left quadrant.
To combine this information into a single edge image, we simply zero the
LL subband of the transform, compute the inverse transform, and take the
absolute value.
The next Figure shows the modified transform and resulting edge image.
(a)
(b)
Figure 9.9 (a) transform modified by zeroing the LL subband. (b) resulted edge image
(a)
(b)
Figure 9.10 (a) noisy image. (b) its two-level wavelet transform