Digital Image Fundamentals Human Vision Lights and Electromagnetic spectrum Image Sensing & Acquisition Sampling & Quantization Basic Relationships b/w.

Digital Image Fundamentals

Human Vision

Lights and Electromagnetic spectrum

Image Sensing & Acquisition

Sampling & Quantization

Basic Relationships b/w Pixels

Digital Image Processing 2

Important dates 9/29: Project grouping (2~3 members/group) 10/6: First image processing GUI due!!

OpenCV ImageMagick ImageJ Ximage or ImageX


A Cross Section of the Human Eye

• Iris – 虹膜• Lens – 水晶體• Cornea – 角膜• Sclera – 鞏膜• Choroid – 脈絡膜

• Retina – 視網膜• Fovea – 視乳頭• Ciliary body – 睫狀體


Human Vision

Rods: 108

Shape/form perception Large dynamic range Limited contrast Scotopic (dim-light) vision

Cones 5 X 106

3-channel color perception Photopic (bright-light) vision


Distribution of Rods and Cones in the Retina


Image Formation in the Eye


Range of Subjective Brightness

• Visual system cannot operate over full range of subjective brightness simultaneous

• Via brightness adaptation


Brightness Discrimination

Weber ratio / intensity

• ∆Ic – increment of illumination

discriminable 50% of the time I

• Small ∆Ic/I => good discrimination; otherwise, poor.


Perceived Brightness

Two phenomena

• Undershoot or overshoot around the boundaries; Mach band pattern

• Simultaneous contrast


Optical Illusions


Electromagnetic Spectrum


Image Sensing

• Single sensor

• Sensor strip

• Sensor array


A Simple Image Model

i(x,y) – illumination (from light source) r(x,y) – reflectance of illuminated

surface (reflectivity)

Lambertian surface Looks the same in all directions

Specular (mirror-like) surface Incidence angle = reflectance angle


A Simple Image Model (continued)

f(x,y) = i(x,y) X r(x,y) >= 0

r(x,y) 0.93 white snow 0.01 black velvet

i(x,y) 9000 foot-candle Sun 0.01 foot-candle full moon


Sampling & Quantization


A Digital Image of MXN Array

)1,1()1,1()0,1(

)0,2(

)1,1(

)1,0()2,0(

)1,1()0,1(

)1,0()0,0(

),(

NMfMfMf

Mf

Nf

Nff

ff

ff

yxf


A Digital Image (continued)

• Image Sampling – Spatial-coordinate digitization

• Gray-level Quantization – amplitude digitization

N = size of image = (number of columns) X (number of rows)

G (number of gray levels) = 2k

Disk storage needed = N * ceiling(k/8)


Storage Bits for N and k


Spatial Resolution


Amplitude Quantization


Level of Detail (LOD)

Low level of detail High level of detail


Isopreference


Scaling and Interpolation


Basic Image Topology

Neighbors of a Pixel 4-neighbor and 8-neighbor 4-adjacent and 8-adjacent

Connectivity 4-connectivity 8-connectivity M-connectivity (mixed connectivity)


M-Connectivity

})()({)( (ii)

or ),( (i)

:set foregroundrt adjacent w are and

44

4

VqNpNpNq

pNq

Vm-qp

D


Further Pixel Relationships

Connected Component Labeling Relations, Equivalence, and Transitive

Closure Distance Measures Arithmetic/Logic Operations Mask Operations


Logic Mask Operations


Weighted Mask Operation


Utilizing ALU Parallel Processing

Digital Image Fundamentals Human Vision Lights and Electromagnetic spectrum Image Sensing & Acquisition Sampling & Quantization Basic Relationships b/w.

Documents