Two-dimensional object detection in correlated noise

摘要

In this paper, a procedure for the detection of two-dimensional objects embedded in correlated noise is presented. The procedure developed here is an extension of the dependent noise model of the recent work by Chang and Kurz (Object detection and experimental designs, Comput. Vision Graphics Image Process.40, 147–168 (1987)). The model for the dependent data is based on the two-way analysis, under the analysis of variance (ANOVA), with fixed effects and known correlation matrix. The detection is based on the test of homogeneity and heterogeneity properties of target and background arrays and the combination of both: the global array. For each data array, F-statistics are calculated and tested against tabulated thresholds. For the case of objects with large size, a reduced algorithm in the form of a contrast test statistic is proposed. Results of computer simulations carried out to evaluate the new algorithms are presented.