lecture07

lecture07 - ELEC317 Digital Image Processing Lecture 7...

Info iconThis preview shows pages 1–7. Sign up to view the full content.

View Full Document Right Arrow Icon
ELEC317 Digital Image Processing Lecture 7 Image Analysis and Computer Vision (part 1) Here we would like to extract features from an image, then describe, interpret and understand the image. Steps Involved: 1. Preprocessing -- Restoration (Noise removal, deblurring, etc) -- Enhancement -- Coding 2. Segmentation -- Separate image into components -- May need feature extraction first 3. Feature Extraction 4. Symbolic Representation of Features 5. Classification and Interpretation Intermediate-level processing Problem domain Elements of Image Analysis Results High-level processing Low-level processing Recognition and Interpretation Image Acquisition Representation and Description Segmentation Preprocessing Knowledge Base 1
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
Applications 1. Character Recognition -- Mail Sorting -- Label Reading -- Text Reading etc. 2. Medical Image Analysis -- Cell Counting -- Tumor Detection 3. Industry -- Parts Identification -- Defects / Fault Inspection 4. Robotics -- Machine Vision 5. Forensics ( ) -- Finger-print Matching -- Bank-check Signature Verification 6. Radar Imaging -- Remotely Piloted Vehicles -- Missile Guidance System Image Feature Extraction Features: -- Amplitude -- Histogram Feature -- Edge / Gradient -- Region -- Boundary -- Moments -- Structure -- Shape -- Texture We will discuss each of the above features and how to extract them next. 2
Background image of page 2
Amplitude Feature -- Here the fact that certain object / region of interest has specific pixel ranges is utilized. -- Feature extracted by window slicing. C a b a b Histogram Features -- Here we use the fact that similar objects have similar histogram -- Histogram can also be regarded as pdf of pixel values. Prob { u = x } = ) ( x P u pdf ( x ) Pixels of Number Total x Level Gray with Pixel of Number = x -- Sometimes, histogram is too complicated to use. Thus, instead of pdf of u , we use moments of u : i th Moment: M i = E { u i }= = 1 0 ) ( L x u i x P x i = 1 Î Mean i = 2 Î Mean Squared Value or Average Energy 3
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
i th Central Moment: μ i = E {[ u - E { u }] i }= = 1 0 1 ) ( ) ( L x u i x P m x i = 2 Î Variance i = 3 Î Skewness i = 4 Î u 4 -3: Kurtosis i th Absolute Central Moment: = E {| u - E { u }| i µ ˆ i }= = 1 0 1 ) ( | | L x u i x P m x i = 1 Î Dispersion -- For texture extraction, second-order joint distribution is useful: Prob { u = ) , ( 2 1 2 1 x x P u u 1 = x 1 ; u 2 = x 2 } Pixels of Pairs of Number Total x u x u pixels of Pairs of Number 2 2 1 1 ; = = = Transform Features -- The region of interest is transformed and is inspected in transform domain. e.g. Edge Extraction ˆ 1 U(k,l ) HP Filtered Image ( m,n ) (Mainly edges) HPF Mask Input Image u(m,n ) 4
Background image of page 4
LPF Horizontal HPF Horizontal BPF Horizontal LPF (Vertical allpass) HPF BPF Vertical LPF (Horizontal allpass) Vertical BPF Vertical HPF Edge Detection y f x f -- Edge ÍÎ Boundary Good for: 1. Segmentation 2. Registration (Alignment) 3. Object Identification 5
Background image of page 5

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
-- Edge is location where there is abrupt change in gray-level. e.g.
Background image of page 6
Image of page 7
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 04/14/2010 for the course ELEC 317 taught by Professor Nil during the Spring '02 term at HKUST.

Page1 / 19

lecture07 - ELEC317 Digital Image Processing Lecture 7...

This preview shows document pages 1 - 7. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online