qcchypklt.3

spiht for linux this is used to decod and encode vedio i wich all enjoy

.TH QCCHYPKLT 3 "QCCPACK" ""
.SH NAME
QccHYPklt \- 
data structure
.B QccHYPklt
for spectral KLT transform
.SH SYNOPSIS
.B #include "libQccPack.h"
.sp
.BI "int QccHYPkltInitialize(QccHYPklt *" klt );
.br
.BI "int QccHYPkltAlloc(QccHYPklt *" klt );
.br
.BI "void QccHYPkltFree(QccHYPklt *" klt );
.br
.BI "int QccHYPkltPrint(const QccHYPklt *" klt );
.br
.BI "int QccHYPkltRead(QccHYPklt *" klt );
.br
.BI "int QccHYPkltWrite(const QccHYPklt *" klt );
.SH DESCRIPTION
QccPack provides data structure
.B QccHYPklt
for representing a KLT transform.
.LP
The components of a
.B QccHYPklt
structure are a vector representing the dataset mean and
a matrix representing the transform itself, as determined
by a singular value decomposition of the dataset's covariance
matrix.
.SH "DATA STRUCTURE"
The
.B QccHYPklt
data structure is defined as:
.RS
.nf

typedef struct
{
  QccString filename;
  QccString magic_num;
  int major_version;
  int minor_version;
  int num_bands;
  QccVector mean;
  QccMatrix matrix;
} QccHYPklt;
.fi
.RE
.LP
The fields of
.B QccHYPklt
are as follows:
.TP
.IR filename
The name of the file.
.TP
.IR magic_num ", " major_version ", " minor_version
The magic number and version of the file.
.TP
.IR num_bands
The number of spectral bands the KLT is designed for.
.TP
.IR mean
The mean vector (length =
.IR num_bands ).
.TP
.IR matrix
The transform matrix (size = 
.IR num_bands " x " num_bands ).
.SH "FILE FORMAT"
For reading and writing structures
of type
.BR QccHYPklt ,
QccPack provides the
.B KLT
file format.
This file format starts with an ASCII header followed by
binary data.
The ASCII header consists of magic-number/revision
information
followed by any amount of white space
(space, `\\t' (tab), `\\n' (newline), `\\r' (return)) and/or
comments lines (lines starting with `#').  Following this white space,
additional ASCII
header information is given, separated by blanks and newlines.
Binary data follows this ASCII header information.
.LP
The
.B KLT
file format consists of the following information:
.RS
.sp
.BI KLT X.X
.br
.I "<white space>"
.br
.I num_bands
.br
.I mean
.br
\|.
.br
\|.
.br
\|.
.br
.I matrix
.br
\|.
.br
\|.
.br
\|.
.br
.sp
.RE
where
.B KLT
is the magic number,
.I X.X
is the version number,
.I "<white space>"
is white space and/or 
comment lines, 
.I num_bands
is the number of bands,
.I mean
are the elements of the mean vector,
and
.I matrix
are the elements of the transform matrix.
.I num_bands
is stored in ASCII, while the
.I mean
and
.I matrix
values are stored as binary
(4 bytes, MSB first, see
.BR QccFileWriteDouble (3)).
No white space may exist between the 
.I mean
and
.I matrix
values.
.SH "ROUTINES"
.B QccHYPkltInitialize()
should be called before any use of a
.B QccHYPklt
structure.
.B QccHYPkltInitialize()
initializes the fields of
.I klt
to the following values:
.RS

.IR filename :
.B NULL
string
.br
.IR magic_num :
.B QCCFILTERBANK_MAGICNUM
.br
.IR major_version ", " minor_version :
initialized to output of 
.BR QccGetQccPackVersion (3)
.br
.IR num_bands :
0
.br
.IR mean :
.B NULL
.br
.IR matrix :
.B NULL
.br
.RE
.LP
.B QccHYPkltAlloc()
allocates the
.I mean
and
.I matrix
fields of the KLT.
.IR num_bands
must be set prior to calling
.BR QccHYPkltAlloc() .
.LP
.B QccHYPkltFree()
frees the 
.I mean
and
.I matrix
fields of
.IR klt .
.LP
.B QccHYPkltPrint()
prints the contents of
.I klt
to stdout.
.LP
.BR QccHYPkltRead()
reads a
.BR QccHYPklt
structure from a file;
the filename of the
.BR KLT -format
file to be read must be stored as
.IR klt->filename .
.LP
.BR QccHYPkltWrite()
writes a
.BR QccHYPklt
structure to a file;
the filename of the
.BR KLT -format
file to be written must be stored as
.IR klt->filename .
.SH "THEORY"
The Karhunen-Lo鑦e transform (KLT) is a data-dependent linear
transform that is optimal in the sense that it
packs the most signal energy into a given number of
transform coefficients.
.LP
The KLT is often used for spectral decorrelation in a hyperspectral
image. Let
.I X
be a collection of pixel vectors; i.e.,
.I X
is a
.IR num_bands " x " num_pixels
matrix, where
.I num_bands
is the number of spectral bands in the hyperspectral image.
Let
.I u
be the mean vector of the pixel vectors (a
.IR num_bands " x 1"
vector).
The covariance matrix
.I R
of
.I X
is then
.IR R " = " X * X ^T/ num_pixels " - " u * u ^T,
where the size of 
.I R
is
.IR num_bands " x " num_bands .
The singular value decomposition of
.I R
is
.IR R " = " U * S * V ^T,
where
.I U
is the transform matrix for the KLT (size
.IR num_bands " x " num_bands ).
.LP
The forward KLT transform is applied to the dataset in question
using the transpose of the matrix
.IR U .
Specifically, for a
.IR num_bands " x 1"
pixel vector
.IR x ,
the forward KLT is
.IR y " = " U ^T*( x " - " u ).
The inverse KLT transform uses the matrix
.IR U
directly; i.e.,
.IR x " = " U * y " + " u .
Here,
.IR u
is the mean vector.
.LP
In QccPack,
.BR QccHYPkltTrain (3)
determines the mean vector
.I u
and the transform matrix
.I U
given the set of pixel vectors in a hyperspectral image;
.BR QccHYPkltTrain (3)
uses
.BR QccMatrixSVD (3)
to perform the SVD of the covariance matrix.
For application of the forward and inverse transforms to the 
image cube, use
.BR QccHYPkltTransform (3)
and
.BR QccHYPkltInverseTransform (3),
respectively.
.LP
The KLT is also called the Hotelling transform or
principal component analysis (PCA).
The name PCA is typically used when only a subset of transform coefficients
(i.e., the principal components) is retained; that is,
PCA typically refers to dimensionality reduction in addition
to the spectral decorreltation inherent in the KLT.
.SH "RETURN VALUE"
These routines return 0 on success, and 1 on failure.
.SH "SEE ALSO"
.BR QccHYPkltTrain (3),
.BR QccHYPkltTransform (3),
.BR QccHYPkltInverseTransform (3),
.BR QccMatrixSVD (3),
.BR QccIMGImageCube (3),
.BR QccPackHYP (3),
.BR QccPackIMG (3),
.BR QccPack (3)
.SH AUTHOR
Copyright (C) 1997-2009  James E. Fowler
.\"  The programs herein are free software; you can redistribute them an.or
.\"  modify them under the terms of the GNU General Public License
.\"  as published by the Free Software Foundation; either version 2
.\"  of the License, or (at your option) any later version.
.\"  
.\"  These programs are distributed in the hope that they will be useful,
.\"  but WITHOUT ANY WARRANTY; without even the implied warranty of
.\"  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
.\"  GNU General Public License for more details.
.\"  
.\"  You should have received a copy of the GNU General Public License
.\"  along with these programs; if not, write to the Free Software
.\"  Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.