US 6,983,222 B1
Multi-stage planar stochastic mensuration
Francis J. O'Brien, Jr., Newport, R.I. (US)
Assigned to The United States of America as represented by the Secretary of the Navy, Washington, D.C. (US)
Filed on Jun. 01, 2004, as Appl. No. 10/863,838.
Int. Cl. G06F 101/14 (2006.01); G06F 17/18 (2006.01)
U.S. Cl. 702—181 11 Claims
OG exemplary drawing
 
1. A method for characterizing a plurality of sparse data sets with less than twenty to thirty data points in a two-dimensional Cartesian space as random or nonrandom, said data sets being based on a plurality of measurements of one or more physical phenomena, said method comprising the steps of:
reading in data points for a first data set from said plurality of data sets;
counting said data points to determine a total number N of said data points;
determining an amplitude range of said data points;
selecting a false alarm rate whereby a rate is provided for which random data will produce a false alarm;
performing a first test comprising utilizing a nonparametric discrete probability distribution for initially classifying said first data set as random or nonrandom as per said first test; and
performing a second test comprising:
partitioning x and y axes of said two-dimensional Cartesian space comprising integer partitioned spaces with unitary intervals based on a maximum range of x values of said data points and a maximum range of said y values of said data points,
forming a second grid with a plurality of partitions based on said unitary intervals,
designating each partition as zero if that partition contains no data points and as one if that partition contains a data point,
forming a sequence of zeros and ones by sequentially looking at each row of said second grid and whether each partition is designated as zero or one,
determining a number of runs r in said sequence wherein each run is a homogenous stream of zeros or ones followed by a different stream of zeros or one's wherein a total number of ones is n1 and a total number of zeros is n2,
computing a Gaussian statistic Z and probability p from n1 and n2, and
initially classifying said first data set as nonrandom if p is greater than said false alarm rate and random if p is greater than said false alarm rate as per said second test; and
utilizing said steps of initially classifying said first data set as per said first test and as per said second test to finally classify said first data set as random or nonrandom.