tailieunhanh - Báo cáo hóa học: " A Statistical Detection of an Anomaly from a Few Noisy Tomographic Projections"

Tuyển tập báo cáo các nghiên cứu khoa học quốc tế ngành hóa học dành cho các bạn yêu hóa học tham khảo đề tài: A Statistical Detection of an Anomaly from a Few Noisy Tomographic Projections | EURASIP Journal on Applied Signal Processing 2005 14 2215-2228 2005 Hindawi Publishing Corporation A Statistical Detection of an Anomaly from a Few Noisy Tomographic Projections Lionel Fillatre ISTIT FRE CNRS 2732 Universite de Technologic de Troyes 12 rue Marie Curie BP 2060 10010 Troyes Cedex France Email fillatrl@ Igor Nikiforov ISTIT FRE CNRS 2732 Universite de Technologie de Troyes 12 rue Marie Curie BP 2060 10010 Troyes Cedex France Email nikiforov@ Received 1 January 2004 Revised 19 November 2004 The problem of detecting an anomaly target from a very limited number of noisy tomographic projections is addressed from the statistical point of view. The imaged object is composed of an environment considered as a nuisance parameter with a possibly hidden anomaly target. The GLR test is used to solve the problem. When the projection linearly depends on the nuisance parameters the GLR test coincides with an optimal statistical invariant test. Keywords and phrases statistical hypotheses testing non linear parametric model nuisance parameter invariant tests missing observations computerized tomography. 1. INTRODUCTION Computerized tomography CT is a technique for reconstructing an object from its projections that are essentially the collections of line integrals of the attenuation scalar field at some set of orientations. The noninvasive nature of tomography has made it very useful for a variety of applications including medical imaging quantitative nondestructive testing object recognition and biomedical system monitoring among others 1 2 3 . In certain practical applications like baggage X-ray scanning or nondestructive testing only a few projections are available and hence a perfect reconstruction of the scene is impossible 2 . The detection of an anomaly target from projections has been already studied in 4 where the authors estimate the unknown location of a parameterized object by maximum likelihood ML estimation. Recently in 5 it is proposed to .