View source for pca

{{DavinciWiki_Function|
Principal Component Analysis
 
Compute the principal components of an N band real data image

NOTE: for the old 3 band pca version see [[pca_byte|pca_byte()]].  This version replaced the old byte version on 2-9-2011
|
Arguments: A numeric array and ignore value
 
Return Value: The rotated data
|
[[Syntax]]: pca(image [, ignore&#61;FLOAT][, struct&#61;BOOL])
 
'image' - A N-band image or data array

'ignore' - the ignore value for the black space of an image

'struct' - return the eigen vectors/values and covariance matrix
|
 dv> a
 600x600x3 array of byte, bip format [1,080,000 bytes]
 
 dv> b&#61;pca(a)
   data: 600x600x3 array of double, bip format [8,640,000 bytes]
|
Original 3-band Image:
 
[[Image:unfiltered.png|250px]]
 
Three principal components:
 
[[Image:comp1.png|250px]] [[Image:comp2.png|250px]] [[Image:comp3.png|250px]]
 
Original 6-band Image:
 
[[Image:NYC-Band_Comparison.PNG|300px]]
 
Six principal components:
 
[[Image:NYC-PCA-C1.PNG|300px]] [[Image:NYC-PCA-C2.PNG|300px]] [[Image:NYC-PCA-C3.PNG|300px]] [[Image:NYC-PCA-C4.PNG|300px]] [[Image:NYC-PCA-C5.PNG|300px]] [[Image:NYC-PCA-C6.PNG|300px]]

|
*[[mxm]]
*[[eigen]]
*[[covar]]
|
*[[eigen2]]
*[[kt1]]
*[[kt2]]
*[[kt3]]
*[[kt4]]
*[[kt5]]
*[[kt6]]
*[[pca2]]
*[[pca_byte]]
|
*general_science.dvrc
|
1.02
}}
[[category:Functions]]
[[category:User Defined Functions]]
[[category:Matrix]]