NEW NONPARAMETRIC STATISTICAL TEST FOR PROBLEMS WITH THREE SAMPLES, WHICH IS MORE EFFECTIVE THAN THE WHITNEY TEST
G. I. Salov
Institute of Computational Mathematics and Mathematical Geophysics, Siberian Branch, Russian Academy of Sciences, pr. Akademika Lavrent’eva 6, Novosibirsk, 630090 Russia
Keywords: three samples, homogeneity test, nonparametric test, detection of objects, noisy image
Abstract
A new nonparametric statistical test for testing the hypothesis of homogeneity of three samples against an alternative hypothesis, which implies that random variables of one of these samples tend to be stochastically greater than random variables of each of the other two samples separately, is proposed. The known Whitney test is equivalent to a particular case of the new test. Application of these tests in a real difficult problem of detection of a specified extended object in a noisy image in a situation with possible appearance of another “interfering” object of a greater size is considered.
|
