Principal Component Analysis PCA Markku HautaKasari Jussi Parkkinen

  • Slides: 23
Download presentation
Principal Component Analysis (PCA) Markku Hauta-Kasari Jussi Parkkinen University of Joensuu Color Group FINLAND

Principal Component Analysis (PCA) Markku Hauta-Kasari Jussi Parkkinen University of Joensuu Color Group FINLAND

Algorithm 1. 2. 3. 4. Calculate correlation matrix for the data set Calculate eigenvalues

Algorithm 1. 2. 3. 4. Calculate correlation matrix for the data set Calculate eigenvalues and eigenvectors for the correlation matrix Select eigenvectors corresponding to the largest eigenvalues as a new basis for the data set Calculate principal components between eigenvectors and the data set Applications: compression, pattern recognition etc. Frequently used for spectral image data

An example on calculating the PCA Example on PCA for the data set of

An example on calculating the PCA Example on PCA for the data set of 2 dimensional vectors

Spectral Images & Principal Component Analysis (PCA) spectrum 1. calculating correlation matrix and eigenvectors,

Spectral Images & Principal Component Analysis (PCA) spectrum 1. calculating correlation matrix and eigenvectors, selecting base vectors 2 -D spectral image Matrix B needs to be transposed 2. calculating inner product images 3. reconstructing spectral image

Illustration of PCA base vectors B original spectral image S 1. 2. inner product

Illustration of PCA base vectors B original spectral image S 1. 2. inner product images P

Illustration of PCA base vectors B ~ reconstructed spectral image S 3. inner product

Illustration of PCA base vectors B ~ reconstructed spectral image S 3. inner product images P

An example of the PCA for the spectral image of printed product

An example of the PCA for the spectral image of printed product

Eigenvectors in case of the pure radiance data

Eigenvectors in case of the pure radiance data

Eigenvectors in case, in which the data is multiplied by a spectral sensitivity of

Eigenvectors in case, in which the data is multiplied by a spectral sensitivity of a human eye and by the used light source

Inner-product images between the spectral images and eigenvectors Inner-product images between eigenvectors and data,

Inner-product images between the spectral images and eigenvectors Inner-product images between eigenvectors and data, which is multiplied by a spectral sensitivity of a human eye and the used light source

An example of the PCA for the spectral image measured from the PDA-display

An example of the PCA for the spectral image measured from the PDA-display

Sample spectral images of a PDA-display

Sample spectral images of a PDA-display

Light sources PDA display and external illumination

Light sources PDA display and external illumination

Inner-product images between the spectral images and eigenvectors Corresponding eigenvectors

Inner-product images between the spectral images and eigenvectors Corresponding eigenvectors

Inner-product images between eigenvectors and data, which is multiplied by a spectral sensitivity of

Inner-product images between eigenvectors and data, which is multiplied by a spectral sensitivity of a human eye and by the used light source Corresponding eigenvectors

Inner-product images between the spectral images and eigenvectors Corresponding eigenvectors

Inner-product images between the spectral images and eigenvectors Corresponding eigenvectors

Inner-product images between eigenvectors and data, which is multiplied by a spectral sensitivity of

Inner-product images between eigenvectors and data, which is multiplied by a spectral sensitivity of a human eye and by the used light source Corresponding eigenvectors

Summary PCA can be effectively used for spectral image compression PCA can be used

Summary PCA can be effectively used for spectral image compression PCA can be used for pattern recognition tasks Exercises on implementing the PCA

SPECTRAL COLOR IMAGING Jussi Parkkinen Markku HautaKasari IPCVSPECTRAL COLOR IMAGING Jussi Parkkinen Markku HautaKasari IPCV
Principal Component Analysis PCA Principal Component Analysis PCAPrincipal Component Analysis PCA Principal Component Analysis PCA
6 Principal Component Analysis Principal component analysis PCA6 Principal Component Analysis Principal component analysis PCA
Principal Component Analysis PCA Introduction Principal component analysisPrincipal Component Analysis PCA Introduction Principal component analysis
6 Principal Component Analysis Principal component analysis PCA6 Principal Component Analysis Principal component analysis PCA
Principal Components Analysis PCA Principal Components Analysis PCAPrincipal Components Analysis PCA Principal Components Analysis PCA
Spectral Color Analysis Jussi Parkkinen University of JoensuuSpectral Color Analysis Jussi Parkkinen University of Joensuu
EE2402009 PCA EE2402009 PCA Principal Component Analysis EE2402009EE2402009 PCA EE2402009 PCA Principal Component Analysis EE2402009
Component Component Component Component Business Components Component basedComponent Component Component Component Business Components Component based
Principle Component Analysis PCA Networks 5 8 PCAPrinciple Component Analysis PCA Networks 5 8 PCA
Multivariate statistics PCA principal component analysis Correspondence analysisMultivariate statistics PCA principal component analysis Correspondence analysis
Multivariate statistics PCA principal component analysis Correspondence analysisMultivariate statistics PCA principal component analysis Correspondence analysis
Multivariate statistics PCA principal component analysis Correspondence analysisMultivariate statistics PCA principal component analysis Correspondence analysis
Principal Component Analysis What is it Principal ComponentPrincipal Component Analysis What is it Principal Component
Principal Nested Spheres Backwards PCA Backwards PCA BackwardsPrincipal Nested Spheres Backwards PCA Backwards PCA Backwards
Geenit ja aatteet Jussi Tammisola kasvinjalostuksen dosentti jussiGeenit ja aatteet Jussi Tammisola kasvinjalostuksen dosentti jussi
Neuvostohallinnon synty ja evoluutio Markku Kangaspuro Markku kangaspurohelsinkiNeuvostohallinnon synty ja evoluutio Markku Kangaspuro Markku kangaspurohelsinki
ALAKOULUN USKONTO Markku Salakka Markku salakkaoulu fi KTKALAKOULUN USKONTO Markku Salakka Markku salakkaoulu fi KTK
Principal Component Analysis Mark Stamp PCA 1 IntroPrincipal Component Analysis Mark Stamp PCA 1 Intro
Multivariate methods Principal component analysis PCA The ideaMultivariate methods Principal component analysis PCA The idea
Principal Component Analysis PCA Learning Representations Dimensionality ReductionPrincipal Component Analysis PCA Learning Representations Dimensionality Reduction
Principal Component Analysis PCA or Empirical Orthogonal FunctionsPrincipal Component Analysis PCA or Empirical Orthogonal Functions