SIGMERG

Description

Merges two class signatures to create a new class signature. The user can specify a gray-level value, threshold, and 'a priori' probability for the new signature class.

Parameters

Name	Type	Length	Value range
InputSIG1: Input class signature segment #1 *	SIG port	1 - 1
InputSIG2: Input class signature segment #2 *	SIG port	1 - 1
OutputSIG: Output merged class signature segment *	SIG port	1 - 1
Gray Level Value	Integer	0 - 1	1 - 254
Gaussian Threshold	Float	0 - 1
Class Bias	Float	0 - 1
Signature Segment Name	String	0 - 1
Signature Segment Description	String	0 - 1

* Required parameter

Parameter descriptions

InputSIG1: Input class signature segment #1

Specifies the first of two class signature segments (type 121) to merge.

Specified input segments (1 and 2) must have the same number and set of channels.

InputSIG2: Input class signature segment #2

Specifies the second of two class signature segments (type 121) to merge.

Specified input segments (1 and 2) must have the same number and set of channels.

OutputSIG: Output merged class signature segment

Specifies the output segment to receive the merged class signature.

Gray Level Value

Specifies an integer value (1-254) that represents the class in the output classification theme map created by MLC.

Values of 0 and 255 are invalid, as they will be assigned to the null and multiply-classified pixels of the output theme map, respectively.

If this parameter is not specified, the value is inherited from the value held in the first input signature segment (DSIG1).

Gaussian Threshold

Specifies the Gaussian feature-space threshold. The greater this value is, the larger the hyperellipsoid or parallelepiped in feature space.

This value is a REAL number in units of standard deviation.

If this parameter is not specified, the value is inherited from the value held in the first input signature segment (DSIG1).

Class Bias

Specifies the BIAS, or relative 'a priori' probabilities for the class. If all classes have equal bias, their 'a priori' probabilities are equal.

This parameter "biases" the maximum likelihood classifier's assignment of a class to a point that lies in overlapping classes in feature-space. This value is a REAL number.

If this parameter is not specified, the value is inherited from the value held in the first input signature segment (DSIG1).

Signature Segment Name

Specifies a name (up to 8 characters) for the output segment.

If this parameter is not specified, the value is inherited from the value held in the first input signature segment (DSIG1).

Signature Segment Description

Describes (in up to 64 characters) the contents or origins of the output data.

If this parameter is not specified, the value is inherited from the value held in the first input signature segment (DSIG1).

Details

SIGMERG merges two existing signature class segments (DSIG) to create a new signature segment.

The merged signature's value (VALU), threshold (THRS), bias (BIAS), segment name (DBSN), and segment description (DBSD) may be specified by the user. Any values not specified will be inherited from the corresponding values of the first input signature segment.

Use SIGMERG for the aggregation of class signature segments, rather than for merging training area bitmaps and creating signature segments. Certain limitations, however, should be recognized. First, the two signature segments used for merging must have been created using the same number and set of input channels. Second, if the two signatures used as input were created from overlapping training area bitmaps, the results will be erroneous, because the overlap area will be sampled twice.