COLOR IMAGE RETRIEVAL BASED ON NON -PARAMETRIC STATISTICAL TESTS OF HYPOTHESIS

摘要: A novel method for color image retrieval based on statistical non-parametric tests such as two-sample Wald Test equality of variance and Man-Whitney U test is proposed in this paper. The the deviation, i.e. distance terms between query target images; if images pass test, then it proceeded to spectrum energy, mean values two otherwise, dropped. If inferred that belong same class, both are same; assumed different classes, different. obtained statistic indexed ascending order corresponds least value identified or similar images. Here, either treated sample; other population. Also, some features Coefficient Variation, Skewness, Kurtosis, Variance, Spectrum Energy compared color-wise. robust scaling rotation, since adjusts itself treats sample other. results comparable with existing methods. Keywords—Variance, mean, image, tests.