bildanalyse this course image analysis - uni … · ¥image analysis ¥image understanding...

6
Bildanalyse Bildanalyse Image Analysis Image Analysis Paul Bons Paul Bons AG Strukturgeologie AG Strukturgeologie AB Mineralogie und Geodynamik AB Mineralogie und Geodynamik lderlinstr lderlinstr . 16 . 16 E-mail: [email protected] E-mail: [email protected] URL: URL: http://structural-geology.info http://structural-geology.info 0160-5515482 0160-5515482 This course Overview of techniques for Image processing Image analysis Image understanding Focussing on digital methods Software to use: ImageJ Nat. Inst. Health (USA) http:// rsb .info. nih . gov / ij / Processing, analysis & understanding Processing Manipulation of (raw) image Enhance image, reduce noise, etc. Prepare for analysis Enhance contrast Processing, analysis & understanding Analysis Extract data from image Measurement, segmentation, etc. Segmentation

Upload: dinhlien

Post on 20-Aug-2018

226 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Bildanalyse This course Image Analysis - uni … · ¥Image analysis ¥Image understanding ¥Focussing on digital methods ¥Software to use: ¥ImageJ ¥Nat. Inst. Health (USA)

BildanalyseBildanalyse

Image AnalysisImage Analysis

Paul BonsPaul BonsAG StrukturgeologieAG Strukturgeologie

AB Mineralogie und GeodynamikAB Mineralogie und Geodynamik

HHöölderlinstrlderlinstr. 16. 16

E-mail: [email protected]: [email protected]

URL: URL: http://structural-geology.infohttp://structural-geology.info

0160-55154820160-5515482

This course

• Overview of techniques for

• Image processing

• Image analysis

• Image understanding

• Focussing on digital methods

• Software to use:

• ImageJ

• Nat. Inst. Health (USA)

• http://rsb.info.nih.gov/ij/

Processing, analysis & understanding

• Processing

• Manipulation of (raw) image• Enhance image, reduce noise, etc.

• Prepare for analysis

Enhance contrast

Processing, analysis & understanding

• Analysis

• Extract data from image• Measurement, segmentation, etc.

Segmentation

Page 2: Bildanalyse This course Image Analysis - uni … · ¥Image analysis ¥Image understanding ¥Focussing on digital methods ¥Software to use: ¥ImageJ ¥Nat. Inst. Health (USA)

Processing, analysis & understanding

• Understanding

• Convert data to information• Interpretation, classification of data

• Visualisation of data

Urban areas

This lecture: introduction

• What is an image?

• Basics of digital images

• Introduction to image operations

What is an image?

• An image is a visual projection or map of something(in the real world) onto some medium

• Projection method

• Photo camera

• Electron microscope

• Radar

• Satellite

• Scanner

• Medium

• Celluloid film

• Chip, computer RAM

• Photographic paper

Real world image

camera

Digital image acquisition

• Classical image acquisition captures three visiblelight bands

• Red

• Green

• Blue

Page 3: Bildanalyse This course Image Analysis - uni … · ¥Image analysis ¥Image understanding ¥Focussing on digital methods ¥Software to use: ¥ImageJ ¥Nat. Inst. Health (USA)

Other acquisition methods

• Image acquisition not limited to 3 bands

• Electon microscopy: only one band

• Intensity of secondary electron emission

• Multi-spectral cameras: many bands

• For example ASTER satellite images:• 4 visible light bands

• 8 short-wavelength infrared bands

• However, the human eye can only see colour imagesmade of the three basic colours

Digital image = intensity map

• Light intensity is mapped onto a square grid or arrayof pixels

x

y

pixel3-dimensional "images": voxel

Number of pixels determinesresolution

• More pixels:

• Higher resolution (better quality)

• Bigger size• More memory space

• Slower calculations

128 x 128

49 kbyte

64 x 64

12 kbyte

32 x 32

3 kbyte

16 x 16

768 byte

8 x 8

192 byte

Monochrome (grey-scale) image

• Each pixel has a single value

• Represents the brightness orlight intensity at that position

• Standard convention

• Black = 0 (0% intensity)

• White = 255 (100% intensity)

• 256 possible intensity values

= 1 byte (256=28)X = 40

Y = 228Intensity = 168 (66%)

Page 4: Bildanalyse This course Image Analysis - uni … · ¥Image analysis ¥Image understanding ¥Focussing on digital methods ¥Software to use: ¥ImageJ ¥Nat. Inst. Health (USA)

Number of grey levels

• The human eye is not verysensitive to grey levels

• We can hardly distinguish morethan 16 grey levels

256 levels 32 levels 16 levels

8 levels

Clear quality loss

Band width

• Number of grey levels determinesnumber of divisions between blackand white

• However, what is "black" and whatis "white"?

Imagebrightness

255

0

Saturatedwhite

Saturatedblack True brightness

BAND WIDTH Saturation

Colour images

• The human eye has three sensitivity peaks

• RED (!600 nm wavelength)

• GREEN (!550 nm wavelength)

• BLUE (!440 nm wavelength)

• Standard colour images are therefore RGB

• Alternative CMY

• Cyan-Magenta-Yellow

RED

GREENBLUE

YELLOW

CYAN

MAGENTA

Different bands, same result

colour red green blue

= + +

= + +

cyan magenta yellowPer pixel

3 byte 1 byte 1 byte 1 byte= + +

Page 5: Bildanalyse This course Image Analysis - uni … · ¥Image analysis ¥Image understanding ¥Focussing on digital methods ¥Software to use: ¥ImageJ ¥Nat. Inst. Health (USA)

HSI: Hue-Saturation-Intensity

• Hue = colour (deviation from red)

• Saturation = colour intensity (deviation from grey)

• Intensity = brightness (same as grey-scale image)

• From RGB-colour image to grey-scale image

• Intensity = (R+G+B) / 3

colour hue saturation intensity

= + +

Combining other bands

band 1 band 2 band 3

Short-wavelengthinfrared(>0.72 µm)

• Some sensors recordmore than 3 bands

• Any three bands canbe combined to makea false-colour image

Band 1+2+3

Image operations

• Image operations are operations that convert oneimage to another

• Each pixel in the image is changed to a new value(monochrome) or more values (colour)

• Example: contrast enhancement

3 types of single image operations

• Point

• New value at (x,y)

depends only on original

value at (x,y)

• Local

• New value at (x,y)

depends on original

values around (x,y)

• Global

• New value at (x,y)

depends on all values in

original imageoriginal new

Page 6: Bildanalyse This course Image Analysis - uni … · ¥Image analysis ¥Image understanding ¥Focussing on digital methods ¥Software to use: ¥ImageJ ¥Nat. Inst. Health (USA)

Operations can also involve multipleimages

• All arithmetic operations are possible on two or moreimages

• V(x,y,A) + V(x,y,B) = V(x,y,C)

• V(x,y,A) - V(x,y,B) = V(x,y,C)

• V(x,y,A) x V(x,y,B) = V(x,y,C)

• Etc.

+ =

Image A Image B Image C

End of lecture 1

• If you have a computer/laptop

• Get ImageJ (it's for free!)

• http://rsb.info.nih.gov/ij/

• Play with the program

• Make sure you have an account on the CIP-Poolcomputers

• Register with assistant

• Get a key-card to have access to room at any time