Digital Image Processing

Download Report

Transcript Digital Image Processing

Lecturer: Conf. dr. ing. Mihaela GORDAN Communications Department e-mail: [email protected]

Office phone: 0264-401309 Office address: Multimedia (CTMED) laboratory, Str. C. Daicoviciu Nr. 15

Digital Image Processing

Lecture notes – fall 2010

Digital Image Processing Lecture 1 – Introductory

Lecture 1

• Introduction • Course description • Examination grade information

Digital Image Processing

Introduction (1)

Lecture 1 – Introductory • •

Digital image processing: deals with digital images = digital representation of the visual scenes

Note that: visual perception can be static (scene content unchanged in time) or dynamic (scene content changes in time); the latest case =

video sequence;

Typically, visual scene = a static image, a “snap shot” tries to:

“implement” in digital (algorithmic) form various human vision

processes => image analysis & understanding, pattern recognition

“improve” image appearance for human visualization => image

enhancement, de-noising;

BASIC IMAGE PROCESSING

store and transmit images efficiently => image compression

Digital Image Processing

Introduction (2)

Lecture 1 – Introductory

Applications of digital image processing?

… virtually, everywhere!

Industry: inspection/sorting; manufacturing (robot vision)

Environment: strategic surveillance (hydro-dams, forests, forest fires, mine galleries) by surveillance cameras, autonomous robots

Medicine: medical imaging (ultrasound, MRI, CT, visible)

Culture: digital libraries; cultural heritage preservation (storage, restoration, analysis – indexing)

Television: broadcasting, video editing, efficient storage

Education & tourism: multi-modal, intelligent human-computer interfaces, with emotion recognition components

Security/authentication

… etc…

(iris recognition, signature verification)

Digital Image Processing

Introduction (3)

Lecture 1 – Introductory •

Industrial inspection (industrial vision systems):

Digital Image Processing

Introduction (4)

Lecture 1 – Introductory

Water sources inspection:

Environment surveillance/monitoring:

Forest fire monitoring Hydro sites surveillance

Digital Image Processing Lecture 1 – Introductory

Introduction (5)

Medical imaging applications:

Color image segmentation & Cells counting Ultrasound image analysis/quantification

Digital Image Processing Lecture 1 – Introductory

Course description (1)

… Obviously, digital image processing is a very wide field, sooo… …What will we study in 1 semester…?

Just the basics you need to develop & implement image

processing & analysis algorithms from all the categories above!

Simplification:

-

only grey level images

-

only basic processing methods, without their combination

Digital Image Processing Lecture 1 – Introductory

Course description (2)

Course chapters:

I.

Grey level digital image representation. Basic math concepts for digital image processing algorithms II.

Grey level image digitization: II. 1. Image sampling II. 2. Image quantization III. Image transforms: digital image representation in frequency domains; applications: noise filtering, compression, recognition III. 1. Basic properties III. 2. Sinusoidal transforms III. 3. Rectangular transforms III. 4. Eigenvector-based transforms

Digital Image Processing Lecture 1 – Introductory

Course description (3)

IV. Image enhancement: IV. 1. Point operations IV. 2. Grey level histogram; histogram-based enhancement IV. 3. Spatial operations IV. 4. Transform-based operations IV. 5. Color image enhancement & pseudo-coloring V. Image analysis & understanding: V.1. Regions of interest; features; feature extraction V. 2. Edge detection, boundary extraction & representation V. 3. Regions detection, extraction & representation V. 4. Binary object structure analysis & representation: median axis transforms; binary morphology

Digital Image Processing Lecture 1 – Introductory

Course description (4)

V. 5. Shape descriptors V. 6. Texture representation; texture descriptors V. 7. Region-based image segmentation VI. Image compression & coding: VI. 1. Introduction VI. 2. Pixel coding VI. 3. Predictive coding of still images VI. 4. Transform coding of still images VI. 5. Video sequence (inter-frame) coding

… all with practical examples given – in the lectures & lab!

Digital Image Processing Lecture 1 – Introductory

Examination grade information

The grade components:

1) Written test – quiz: => max. 3.5 pts - 6 questions from theory - 6 questions from problems/exercises 2) Written test – classic: => max. 6.5 pts - 5 short theoretic subjects (max. ½ page answer) - 5 problems/exercises => Written test grade T=1…10 3) Laboratory work evaluation: => grade L=1…10 4) Lecture participation/discussions: => grade LD=1…10 5) Project evaluation: => grade P=1…10 ____________________________________________________________________ The grade = 0.75(0.7T+0.2L+0.1LD)+0.25P

To pass: must have T≥ 4.5, L≥ 5.

Digital Image Processing

References

A) Lecture:

A.Vlaicu –

Prelucrarea imaginilor digitale

. Editura Microinformatica, Cluj N., 1997 Lecture slides – available online

B) Laboratory:

Will be soon available online (as pdf); also online images, some sample applications/code

C) Exercises, written test samples:

Available online The official DIP course site: http://193.226.17.10/sites/pni Lecture 1 – Introductory

Digital Image Processing Lecture 1 – Introductory

Mathematical Representation of Grey Scale Digital Images (1)

Def.: Grey scale image

= visual representation of a finite size 2-D scene, in which the scene is represented, in each spatial position (x,y), by its

brightness

(=grey level value, lightness): - minimum brightness (0) = black; - maximum brightness (L Max )= white.

=> mathematically: let the physical dimensions of the scene be H f height and W f – width; e.g., H f =25cm; W f – =4cm; the scene origin = upper left corner => the space-continuous grey level image is described by the

brightness spatial function:

f:[0;W f )×[0;H f )→[0;L Max ], f(x,y)=the brightness of the scene in the spatial position (x,y)

Digital Image Processing Lecture 1 – Introductory

Mathematical Representation of Grey

(0,0)

Scale Digital Images (2)

δx (0,0) x δy H f

Note:

y W f Continuous grey level scene Scene digitization: discretization of the spatial positions The brightness information is the most important in the scene; it is perceived by a special type of photoreceptors (the rods) in the HVS; the perception of brightness makes possible the orientation at low light (illumination) levels