1 Lecture 1 1 Image Processing Eng. Ahmed H. Abo absa E-mail: [email protected]@up.edu.ps.
Jan 02, 2016
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
2
Lecture 1
What is Digital Image Processing
Processing digital images by means of a digital computer.
A digital image can be modeled as a two dimensional function , ,where x and y are spatial coordinates, and the value of the function is the intensity or gray level of the image at that point.
),( yxf
3
Lecture 1
What is Digital Image Processing
A digital image is composed of a finite number of elements, each of which has a particular location and value. These elements are referred to as picture elements, pixels, and pels.
4
Lecture 1
Digital Image Processing
Image Enhancement
Image Restoration
Image Understanding (or Computer Vision)
Image Coding (or Image Data Compression)
5
Lecture 1
Image Enhancement
Goal to accentuate certain image features for subsequent
analysis or for image display
Input : image Output : image
6
Lecture 1
Image Enhancement
Techniques Contrast enhancement histogram equalization pseudo coloring noise filtering edge sharpening smoothing
Applications processing of remote-sensed image via satellite radar, SAR, Ultrasonic image processing
7
Lecture 1
Image Restoration
Goal to remove or minimize known/unknown degradations in
image
Input : image Output : image
8
Lecture 1
Image Restoration
Techniques De-blurring noise filtering correction of geometric distortion inverse filtering Least mean square(Wiener) filtering
Applications remote-sensed image processing noise cancellation
9
Lecture 1
Image Understanding
Goal to interpret or describe the meaning contained in the
imageInput : image Output : interpretation(description)
““ME”ME”
““circle”circle”
10
Lecture 1
Image Understanding
Techniques boundary descriptor regional descriptor relational descriptor
Applications character recognition automatic inspection of industrial parts ATR(automatic target recognition) target tracking
11
Lecture 1
Image Data Compression
Goal to reduce the amount of data required to represent
images
Input : image Output : bit-stream data
“010100101100110101001 . . . .”
12
Lecture 1
Image Data Compression
Techniques Error-free coding( or lossless coding) Lossy compression Image Compression Standard
JPEG, H.261, H.263, MPEG-1,2,4 etc
Applications Transmission
teleconferencing ,TV system, remote sensing via satellite Storage
VOD(video on demand), Video CD, DVD(digital video disk), medical imaging, educational and business documents
13
Lecture 1
wavelength (Angstroms)
cosmic rays
gamma rays
X-Rays UV
visible
IR
Electromagnetic Spectrum
1 Å = 1 0 - 1 0
m
- 4 - 2 2 4 6 8 1 0 1 2 1
microwave (SAR)
radio frequency
10 10 10 10 10 10 10 10
The whole electromagnetic spectrum is used by “imagers”
Imaging
14
Lecture 1
From the gigantic…
The Great Wall
(of galaxies)
1 0 2 8
m
Scales of Imaging
15
Lecture 1
video camera
1 m
… to the everyday …
Scales of Imaging
16
Lecture 1
… to the tiny.
1 0 - 6 m
electron microscope
Scales of Imaging
17
Lecture 1
Digital Image Formation
18
Lecture 1
Matrix Representation
H=256
W=256
Divide into 8x8 blocks
169130
173129
170181
170183
179181
182180
179180
179179169132
171130
169183
164182
179180
176179
180179
178178167131
167131
165179
170179
177179
182171
177177
168179169130
165132
166187
163194
176116
15394
153183
160183
19
Lecture 1
Image Resolution
20
Lecture 1
Image Resolution
21
Lecture 1
Images and videos are multi-dimensional (≥ 2 dimensions) signals.
2-D image
Dimension 1
Dimension 2
Dimension 1
Dimension 2
Dimension 3
3-D Image Sequence or video
Dimensionality of Digital Images
22
Lecture 1
The Human Visual System (HVS)
LG N
prim ary visua lcortex
h igher leve l vis ionand cogn ition
righ t eyele ft eye
re tina
lens
cornea
fovea
optic nerve
pup il
visua l axis
re tina re tina
23
Lecture 1
HVS: Foveated Vision
Foveated vision: non-uniform resolution of the visual field, highest at the point of fixation and decreasing rapidly
24
Lecture 1
HVS: Visual Illusion
25
Lecture 1
Find the black dot
HVS: Visual Illusion
26
Lecture 1
What is this?
HVS: Visual Illusion
27
Lecture 1
Which lines are straight?
HVS: Visual Illusion
28
Lecture 1
Color
29
Lecture 1
Color: RGB Cube
30
Lecture 1
Color: RGB Representation
31
Lecture 1
Where Are We?
Imaging? Computer Vision?
Display/Printing?
Digital ImageProcessing
Computer Graphics?
BiologicalVision?
32
Lecture 1
What Do We Do?
Image Processing/Manipulation
Image Coding/ Communication
Image Analysis/Interpretation
Digital ImageProcessing
34
Lecture 1
Image Processing: Image Enhancement
Enhance
35
Lecture 1
Image Processing: Image Denoising
Denoise
36
Lecture 1
Image Processing: Image Deblurring
Deblur
37
Lecture 1
Image Processing: Image Inpainting
38
Lecture 1
Image Processing: Image Stylization
39
Lecture 1
Image Analysis: Edge Detection
40
Lecture 1
Image Analysis: Face Detection
41
Lecture 1
Image Analysis: Image Segmentation
42
Lecture 1
Two deceivingly similar fingerprints of two different people
Image Analysis: Image Matching
43
Lecture 1
Image Coding: Image Compression
compressed bitstream
00111000001001101…
(2428 Bytes)
imageencoder
imagedecoder
original image
262144 Bytes
compression ratio (CR) = 108:1
From [Gonzalez &
Woods]
From [Gonzalez &
Woods]
44
Lecture 1
• Lossless image compression– Information preserving
original image can be exactly recovered– Low compression ratio– JPEG-LS, JBIG …
• Lossy image compression– Lose information
original image can be recovered, but not the same
– High compression ratio– JPEG, JPEG2000 …
Image Coding: Image Compression
45
Lecture 1
JPEG (CR=64) JPEG2000 (CR=64)
discrete cosine transform based wavelet transform based
Image Coding: From JPEG to JPEG 2000
46
Lecture 1
Image Coding: Video Compression
• From static images and image sequences (video)– From 2D to 3D– Strong correlations between frames– Representing motion
• Video compression– Compress each frame independently – Motion-compensated video compression
high compression ratio– MPEG1, MPEG2, MPEG4, H.264 …
47
Lecture 1
Image Quality/Distortion Measures
_ =
M
i
N
jijij yx
MNMAE
1 1
1
Y || X_ = Z
ijx| |ijy ijz_ =For each pixel:
Mean Absolute Error (MAE):
48
Lecture 1
Image Quality/Distortion Measures
M
i
N
jijij yx
MNMSE
1 1
21Mean Squared Error
(MSE):
Peak Signal-to-Noise Ratio (PSNR) in decibel (dB):
M
i
N
jijij yx
MN
L
MSE
LPSNR
1 1
2
2
10
2
10 1log10log10
L: Dynamic range of pixel intensityL = 2B – 1, where B is the number of bits to represent a pixelExamples:
8bits/pixel gray-scale image L = 25512bits/pixel gray-scale image L = 4095
49
Lecture 1
Image Quality/Distortion Measures
original
MAE = 0MSE = 0
PSNR = infinity
noisy image 1
MAE = 7.99MSE = 100
PSNR = 28.1dB
noisy image 2
MAE = 15.9MSE = 394
PSNR = 22.2dB
noisy image 3
MAE = 38.2MSE = 2250
PSNR = 14.6dB
50
Lecture 1
Image Quality/Distortion Measures
_ =
9375.1211690430100120144
1
MAE
Y || X _ = Z
ijx| |ijy ijz_ =
1
3
8
6
6
8
6
11
8
10
8
9
9
7
10
10
2
3
8
6
8
8
7
12
5
9
4
15
9
9
1
11
1
0
0
0
2
0
1
1
3
1
4
6
0
2
9
1
• Example: two 4 x 4, 4bits/pixel image
6875.9411368101690100140144
1
MSE
dBMSE
PSNRB
7.136875.9
15log10
)12(log10
2
10
2
10
Related Documents
