Digital Image Processing (DIP): Introduc6on and Fundamentals

Transcription

1 A Digital Image Processing (DIP): Introduc6on and Fundamentals DIP: Introduc-on and Fundamentals I. Origins of DIP

2 Origins of DIP I.1. Newspaper Industry (1920s) Digital picture produced in 1921 Origins of DIP I.2. Space Program (1960s) 3 First picture of the moon by a US spacecra@ in

3 DIP: Introduc-on and Fundamentals II. II.1. Overview 6 Applica6ons in many fields: medicine, geography, physics, astronomy, defense Examples: character recognieon, fingerprint recognieon, target recognieon, weather prediceon, crop assessment, deteceon of bone fractures, deteceon of brain tumors, study of high-energy plasmas

4 7 II.2a. Imaging in the Electromagne6c Spectrum INFRARED RADIO WAVES Energy E Frequency ν MICROWAVES 600 nm 700 nm VISIBLE Wavelength λ λν = c, E = hν 500 nm HARD X-RAYS ULTRAVIOLET SOFT X-RAYS GAMMA RAYS 400 nm 8 Gamma-Ray II.2b. Imaging in the Electromagne6c Spectrum Bone scan (to localize infeceons and tumors)

5 9 X-Ray II.2c. Imaging in the Electromagne6c Spectrum Image of a circuit board (to detect flaws) 10 Ultraviolet II.2d. Imaging in the Electromagne6c Spectrum Fluorescence microscope images of corn (to find if corn is infected by smut)

6 11 Visual Spectrum II.2e. Imaging in the Electromagne6c Spectrum Image of paper currency (to track and idenefy bills) 12 Infrared II.2f. Imaging in the Electromagne6c Spectrum Satellite image of North America (global inventory of human serlements)

7 13 Radio-Band II.2g. Imaging in the Electromagne6c Spectrum MRI image of a human knee (diagnosis of sport-related injuries) nm II.2h. Imaging in the Electromagne6c Spectrum Panchroma6c

8 15 II.2i. Imaging in the Electromagne6c Spectrum nm Colour nm II.2j. Imaging in the Electromagne6c Spectrum

9 17 II.2k. Imaging in the Electromagne6c Spectrum nm False Colour Ultrasound II.3a. Other Imaging Modali6es Image of an unborn baby (to determine the health of his development) 18

10 Synthe6c II.3b. Other Imaging Modali6es Computer-generated image of a human skull (for criminal forensics) DIP: Introduc-on and Fundamentals III. Main Steps in DIP 19

11 21 Main Steps in DIP III.1a. Overview outputs are images outputs are image arributes 4 L O W L E V E L IMAGE PREPROCESSING 3 IMAGE ENHANCEMENT SEGMENTATION 5 2 IMAGE RESTORATION KNOWLEDGE BASE REPRESENTATION AND DESCRIPTION 6 1 IMAGE ACQUISITION L E V E L OBJECT RECOGNITION Main Steps in DIP III.1b. Overview LOW LEVEL: acquisieon and improvement of pictorial informaeon for human interpretaeon and analysis H I G H HIGH LEVEL: autonomous machine percepeon 22

12 23 Main Steps in DIP III.1c. Overview DIP characterized by specific solueons. Techniques that work well in one area can be totally inadequate in another. Actual solueon of a specific problem generally sell requires significant research and development. 24 Main Steps in DIP III.2a. Examples Which step?

13 25 Main Steps in DIP III.2b. Examples Which step? 26 Main Steps in DIP III.2c. Examples Which step?

14 27 Main Steps in DIP III.2d. Examples Which step? DIP: Introduc-on and Fundamentals IV. Image Defini6on and Representa6on

15 Image Defini-on and Representa-on IV.1a. Analog Image funceon f a R 2 [0;+ [ 29 image funceon f a R 3 R funceon f a R 2 R 3 funceon f a R 2 C f a R 2 [0;+ [ is a typical case; f a (x,y) is the intensity of f a at (x,y). Image Defini-on and Representa-on IV.1b. Analog Image 30 z=f a (x,y) y x Consider f a R 2 [0;+ [ z 0 N L N y y 0 M M x x Domain of definieon [0,M]x[0,N] and range [0,L]

16 Image Defini-on and Representa-on IV.1c. Analog Image 31 f a Zero Padding Image Defini-on and Representa-on IV.1d. Analog Image 32 f a Circular Indexing

17 Image Defini-on and Representa-on IV.1e. Analog Image 33 f a Reflected Indexing Image Defini-on and Representa-on IV.2a. Digital Image 34 image funceon f Z funceon f Z 3 Z funceon f Z 2 Z 3 f Z is a typical case : grayscale image ((x,y),f(x,y)) loca6on gray level pixel

18 Image Defini-on and Representa-on IV.2b. Digital Image x. M-1 Consider f Z Assume range included in 0..L-1: typically, L=2 l, i.e., l -bit grayscale image Assume domain of definieon is 0..M-1 x 0..N-1: y... N-1 0 f(x,y) L-1 pixel locaeon f(0,0) f(0,1) f(0,n-1) f(1,0) f(m-1,0) f(m-1,n-1) 35 Image Defini-on and Representa-on IV.2c. Digital Image Consider f Z Assume range included in 0..L-1: typically, L=2 l, i.e., l -bit grayscale image Assume domain of definieon is 0..M-1 x 0..N-1: y 36 f(0,0) f(0,1) f(0,n-1) f(1,0) x pixel f(m-1,0) f(m-1,n-1)

19 37 Image Defini-on and Representa-on IV.3a. Digi6za6on quan6za6on fa R 2 [0;+ [ sampling digi6za6on analog Z 2 [0;+ [ R f Z digital 38 Image Defini-on and Representa-on IV.3b. Digi6za6on Checkerboard effect: sampling 1024

20 IV.3c. Digi6za6on False contouring: 39 quanezaeon Image Defini-on and Representa-on