No Slide Title
Download
Report
Transcript No Slide Title
Digital Image Processing
Chapter 2
Chapter 2: Digital Image Fundamentals
Digital image processing is based on
Mathematical and probabilistic models
Human intuition and analysis
Resolution
Adaptation
2.2 Light and the EM Spectrum
Light is a particular type of EM radiation that
can be seen by human eye.
Green object reflect light with wavelengths primarily
in 500 to 570 nm range.
Chromatic light spans EM spectrum from 0.43
m(violet) to 0.79 m(red).
Radiance: energy in Watt.
Luminance: in lumens(lm), the amount of energy the
observer perceives.
Brightness: subjective description of light perception.
Chapter 2: Digital Image Fundamentals
A single sensor
A sensor strip
In flat bed scanners
A sensor array
In digital cameras
2.4 Image Sampling and Quantization
To acquire digital images from the continuous sensed
data f(x, y):
Digitization in coordinate values: Sampling
Digitization in amplitude values: Quantization.
The resulting image has M rows and N columns as
f (0,0)
f (1,0)
f ( x, y )
f ( M 1,0)
f (0,1)
f (1,1)
f ( M 1,1)
f (0, N 1)
f (1, N 1)
f ( M 1, N 1)
The digitization process requires to determine
the M, N, and L.
Dynamic range: the range of values spanned
by the gray scale, [Lmin, Lmax].
M and N: spatial resolution
L: gray-level resolution
L = 2k. L = gray-level
High dynamic range = high contrast image
The number of bits required to store the
image
b = M N k or
b = N2 k
•Spatial resolution is the smallest discernible detail
in an image.
• line pair width = 2W (line width + space)
• No. of line pairs per unit distance = 1 / 2W
• Resolution is the smallest number of
discernible line pairs per unit distance.
Gray-level resolution refers to the smallest
discernible change in gray level (subjective).
Example of false contouring
2.4.4 Aliasing and Moire Patterns
Band-limited function: the highest frequency is
finite and the function is of unlimited duration
Undersampling – aliasing.
Aliasing frequencies
Sampling rate: the number of samples taken per unit
distance
Reduce high frequency component prior to
sampling.
Moire Pattern is caused by a break-up of the
periodicity, i.e., images are scanned from a
printed page, which consists of periodic ink
dots.
2.4.4 Aliasing and Moire Pattern
Aliasing occurs when a signal is sampled at a less than twice the highest
frequency present in the signal.
The following figure shows if a signal is sampled at regular time intervals
that are slightly less often than once per period of the original signal.
The blue curve is the original signal, and the red dots indicate the sampled values.
資料來源:http://www.wfu.edu/~matthews/misc/DigPhotog/alias/index.html
Aliasing
The red dots are what are recorded, since they represent the
signal values at the times the signal is sampled.
The pattern of the red dots is a terrible representation of the
signal. The red sampled data looks like a sine wave at about
one-tenth the frequency of the original!
This is aliasing.
資料來源:http://www.wfu.edu/~matthews/misc/DigPhotog/alias/index.html
Moire Pattern
2.4.5 Zooming and Shrinking Digital Images
Zooming:
Create a new pixel location.
Assign a gray-level to those new locations
Nearest neighbor interpolation
Pixel replication: a checkboard effect
Bilinear interpolation using four nearest neighbors
v(x’, y’)=ax’+by’+cx’y’+d
where a, b, c, and d are determined from the gray-level of the
four neighbors.
Higher-order non-linear interpolation: using more neighbors for
interpolation
2.4.5 Zooming and Shrinking Digital Images
Proof
2.4.5 Zooming and Shrinking Digital Images
Shrinking:
Direct shrinking causes aliasing
Expansion then Shrinking: blurring the image before shrinking it
and reduce aliasing.
a checkboard effect
2.5 Basic Relations between Pixels
Neighbors of a pixel
Horizontal and vertical neighbors.
Four diagonal neighbors: ND(p)
(x+1, y+1), (x+1, y-1), (x-1, y+1), (x-1, y-1)
4-neighbors of p: N4(p).
8-neighbors of p: N8(p).
(x+1, y), (x-1, y), (x, y+1), (x, y-1)
N8(p) = N4(p) ∪ ND(p)
Adjacency
Connectivity
Regions
Boundary
Relations between Pixels
Adjacency
p
4-adjacency:
N4(p)
p
8-adjacency: N4(p)
∪N (p)
D
• 4-adjacency: if q is in the set N4(p).
• 8-adjacency: if q is in the set N8(p).
• m-adjacency: if
– if q is in the set N4(p), or
– if q is in the set ND(p) and the set N4(p)∩N4(q) has no
pixels whose values are from V.
m-Adjacency
A mixed adjacency combines 4- and 8-adjacency to avoid
the ambigities.
Multiple 8-adjacency
m-adjacency
Two pixels p and q are m-adjacent if
(i) q is in N4(p) or,
(ii)q is in ND(p) and N4(p)∩ N4(q) has no pixel.
Connectivity
Path:
(x0, y0), (x1, y1), …, (xn, yn) where (xi, yi) and (xi+1, yi+1) are
adjacent.
Closed path: (xn, yn) = (x0, y0)
Connectivity:
Two pixels are said connected if they have the same value and there is a
path between them.
If a S is a set of pixels,
For any pixel p in S, the set of pixels that are connected to it is called a
connected component of S.
If S has only one connected component, S is called a connected set.
Regions
R is a region if R is a connected set.
The pixel in the boundary (contour) has at least one 4adjacent neighbor whose value is 0.
Distance measures
Euclidean distance
City-block distance or D4 distance.
D4(p, q)= |x - s | + |y - t |
D8 distance or chessboard distance.
2
2 1 2
2 1 0 1 2
2 1 2
2
D8(p, q)= max (|x - s |, | y - t |)
2 2 2 2 2
2 1 1 1 2
2 1 0 1 2
2 1 1 1 2
2 2 2 2 2