next up previous contents
Next: BLOB analysis methods Up: Extracting features from an Previous: Cutting connections in a

Features from graphs

 

arclength Calculate features from graph
-vdata <nodes> graph nodes
-edata <arcs> graph arcs
-fout <distr> output distribution
-dout <feats> data frame containing calculated features
[-val < -3 | -2 | -1 | gray > ] considered arcs

This command uses a graph defined by nodes and arcs to calculate features based on lengths of arcs. Distribution of arc lengths is saved in field distr and mean, variance, skewness, kurtosis and entropy are stored in data frame feats. Option -val specifies arcs taken into account. If value is zero or positive, only arcs connecting two points with specified color are taken into account. Values -1, -2 and -3 use arcs connecting points that have different color, same color and any color, respectively. Default value is -3.

Example: If nodes and arcs define a graph, following commands plot the length distribution of all arcs.

...
# Calculate length distribution
NDA> arclength -vdata nodes -edata arcs -fout distr \
 -dout features -val -3
# Plot the distribution
NDA> mkgrp g
NDA> ldgrv g -f distr -co black
NDA> show g


next up previous contents
Next: BLOB analysis methods Up: Extracting features from an Previous: Cutting connections in a

Anssi Lensu
Tue Jul 23 11:58:18 EET DST 2002