<< Click to Display Table of Contents >> Navigation:  Help Content > Interferometric Stacking > Post Processing Tools > Time Series Classification > Interferometric Stacking - Post Processing Tools - Time Series Classification - Clustering

Purpose

 

This tool is intended to cluster the deformation time series generated by PS or SBAS processes. It works by grouping nearby time series according to their similarity based on a modified DB Scan algorithm.

Technical Notes

 

Prior to running this tool it is recommended to filter the deformation data using the Filtering tool.

Input Files

Input File List

Single displacement meta file (_meta) or multiple displacement vector files (.shp). Both decomposed and LOS meta files are supported.

Optional Files

Thematic Map

Optional external thematic map.

Parameters - Principal Parameters

 

TS Distance Metric

Metric used to evaluate the similarity of the time series. Euclidean or Cosine available.

 

Maximum Distance

Metric distance threshold to consider two time series similar, expressed as the mean value for each series date after series normalization (if corresponds). Lower values may produce many small clusters, while high values may generate few big clusters

 

Deformation Correlation Length

Maximum distance, expressed in meters, between two points to be considered as neighbors in the TS comparison. This value should be in the order of the dimension of the deformation process present in the scene. Increasing this value will increase the complexity of the process as more points must be analyzed also allowing the creation of giant clusters with low connectivity.

 

L1 threshold

Threshold to consider stable points, those point are not considered for the classification.

 

Minimum Cluster Points

Minimum number of points on a cluster to consider it valid. Clusters with less elements will be mark as unclassified.

 

Min cluster connectivity

Number between 1 and 0 of the minimum mean path length of a cluster to consider it valid.

 

Normalize Time Series

Flag to normalize the time series prior to metric application.

 

Inter-Class Connectivity

Penalization factor for inter thematic region connections. Zero means no inter region connection, 1 no penalization.

Analyze time subset

By setting this flag only the period defined by From and To will be analyzed.

 

Output Files

 

Output Root Name

Name of the output root. This file is mandatory.

 

•Output Root Name_ TS Distance Metric_clustering: Classification map in raster or vectorial format (depending on the input type). Class 0 corresponds to stable points (under the L1 threshold). Class 1 correspond to non-classified points.

•Output Root Name.dot: Clustering graph in graphviz format which can be displayed using software such as Tulip.

•Output Root Name_ TS Distance Metric.shp: Classification vector map. Class 0 corresponds to stable points (under the L1 threshold). Class 1 correspond to non-classified points.

•Output Root Name.dot: Clustering graph in graphviz format which can be displayed using software such as Tulip.

 

General Functions

 

Cancel        

The window will be closed.

 

Help        

Specific help document section.

 

 

Specific Function(s)

 

None.

 

See Also

 

Task, SARscapeBatch object, SARscapeBatch script example

 

References

 

None.