Abstract—By weakening the bigger and strengthening the smaller, gaussianization can enhance the gaussianity of samples and improve performance of subsequent correlation test. Firstly, an explicit definition on gaussianizing filter and a practical method to evaluate the filtering performance are given. Secondly, two typical gaussianizing filters are proposed and studied. One is so-called U-filter, based on the probability density function and its derivate. The other is so-called G-filter, based on the cumulative distribution function and its inverse. Instances with lake trial data are illustrated. Finally, two applications, one in spectrum estimation and the other in Raotest, are discussed.
Index Terms—Gaussianization; Gaussian mixture; Non-Gaussian detection
Wang Pingbo is with the Underwater Acoustic Laboratory, Naval University of Engineering, Wuhan, China (email: email@example.com).
Tang Suofu is with Sonar Equipments Workshop, Naval 702 Maintenance and Repairs Factory, Shanghai, China (email: firstname.lastname@example.org).
Wei Hongkai is with Underwater Acoustic Laboratory, Naval University of Engineering, Wuhan, China (email: email@example.com).
Cai Zhiming is with Underwater Acoustic Laboratory, Naval University of Engineering, Wuhan, China(email: firstname.lastname@example.org).
Cite: Wang Pingbo, Tang Suofu, Wei Hongkai and Cai Zhiming, "Gaussianization for Interference Background," International Journal of Computer Theory and Engineering vol. 2, no. 4, pp. 517-522, 2010.