image processing #2ocw.snu.ac.kr/sites/default/files/note/lecture 11 image... · 2019. 9. 3. ·...
TRANSCRIPT
Image processing #2
고급건설재료학
서울대 건설환경공학부 문주혁 교수
Contexts
• #1. Introduction and Examples
• #2. Basics of Matlab, Image Processing Toolbox
• #3. Segmentation, Edge detection, Transformation
• Matlab code (Image processing toolbox)
• Project introduction
Basics of Matlab
• Numeric types
• Signed integer:
• int8, int16, int32, int64
• Unsigned integer:
• uint8, uint16, uint32, uint64
• Floating point:
• single, double
• Other types
• Logical:
• True, false (1,0)
• Character:
• s = ‘this is a string’
• Variables can be cast to different types:
Basics of Matlab
• Arrays • Order (allocating memory)
Basics of Matlab
• Arrays• Data Structures in Matlab
Basics of Matlab
• Data Structures in Matlab
Basics of Matlab
• Pre-Allocation
Image Processing Toolbox
• Image read and show
Image Processing Toolbox
• Image transformation to black-white (binary) image
Image Processing Toolbox
• Image processing
Image Processing Toolbox
• Image processing
Image Processing Toolbox
• Image processing
Image Processing Toolbox
• Image processing
Image Processing Toolbox
• Image processing
Image Processing Toolbox
• Image processing
Image Processing Toolbox
• Image processing
Image Processing Toolbox
• Region Properties!!!
• Start it over
Image Processing Toolbox
• Region Properties!!!
Image Processing Toolbox
• Region Properties!!! (Use Help! Regionprops)
Image Processing Toolbox
• Image types
• True color (RGB, CMYK etc)
• Grayscale (or gray level, intensity)
• Binary (black & white, bi-level)
Image Processing Toolbox
• Image types
• True color (RGB, CMYK etc)
• Grayscale (or gray level, intensity)
• Binary (black & white, bi-level)
Image Processing Toolbox
• Image types
• True color (RGB, CMYK etc)
• Grayscale (or gray level, intensity)
• Binary (black & white, bi-level)
Image Processing Toolbox
• Ok. Then what is the principle for im2bw? (RGB to Gray to Black & White)
Threshold value를 k라하자.
[1, 2, … , 𝑘]를가지는픽셀들의집합 𝐶0
𝑘 + 1, 2, … , 𝐿 을가지는픽셀들의집합 𝐶1
𝐶0에속할확률 𝑤0 = 𝑤(𝑘)
𝐶1에속할확률 𝑤1 = 1 − 𝑤(𝑘)
𝐶0의평균값 𝜇0 = 𝜇(𝑘)/𝑤(𝑘)
𝐶1의평균값 𝜇1 =𝜇𝑇−𝜇(𝑘)
1−𝑤(𝑘)
𝜎𝐵2 = 𝑤0(𝜇0 − 𝜇𝑇)
2+𝑤1(𝜇1 − 𝜇𝑇)2가최대가되는 𝑘를 1부터 𝐿중에결정
Original 1 threshold
3 thresholds2 thresholds
Project
#1 Particle size analysis of 2D SEM image of superabsorbent polymers
Project
#2 Particle size analysis of 2D SEM image of silica fume
Project
#3 3D pore characteristics analysis of pores in concrete
Project
#4 3D volumetric characteristics analysis of steel fibers in
Ultra-High Performance Fiber-Reinforced Concrete (UHPRFC)
Project
#5 Noise cancellation in video
Project
#6 2D or 3D fiber separation in UHPFRC
Project
#7 Fourier Transformation of TEM image
Lattice images of nanocrystalline regions in C-S-H in OPC specimen 28 d old
C-S-H particle