Statistics I/O utilizes a three-level tree structure. The three levels are:
data_type - The name of the data item. This is a character string
of 26 characters including a null character at the end.
Some examples are: "NAME," "HIST001," "COVARIANCE_MATRIX,"
"POLYGON." Two data_types are reserved by Stat I/O:
"CLASSLIST" and "SITELIST."
Valid items for each level of a statistics file are:
Image Class Site
NBANDS X
NCLASSES X
CONFUSION_MATRIX X
MEAN_VECTOR X X X
COVARIANCE_MATRIX X X X
CORRELATION_MATRIX X X X
HISTOGRAMS X X X
EIGENVALUES X X X
EIGENVECTORS X X X
NPOINTS X X
NSITES X
BOUNDS X
PROBABILITY X
NAME X X
NVERTICES X
POLYGON X
HORIZONTAL_MARK X
shape - A nine element integer array which describes the data
in the record. The first element specifies the data
type. The second element is the number of indices
(dimensions) associated with the data. For example,
if the data is a matrix, then specify 2. The range for
this element is 1 through 7. The remaining seven
elements are the ranges of the indices. Permissible
values are:
= EBYTE: byte data
= EWORD: 16-bit integer
= ELONG: 32-bit integer
= EREAL: single-precision floating point number
= EDOUBLE: double-precision floating point number
= ECOMP: single-precision complex number
= EDCOMP: double-precision complex number
data - The actual statistical data.
An application program can access these routines by linking in the
following object library:
$LASLIB/las.a
This is a list of available Stat I/O Routines: