Adaptive Use of Prior Information in Inverse Problems: An Application to Neutron Depth Profiling
M Levenson, Kevin J. Coakley
A flexible class of Bayesian models is proposed to solve linear inverse problems. The models generalize linear regularization methods such as Tikhonov regularization and are motivated by the ideas of the image restoration model of Johnson et al. (1991). The models allow for the existence of sharp boundaries between different intensity regions in the signal, as well as the incorporation of prior information on the locations of the boundaries. The use of the prior boundary information is adaptive to the data. The models are applied to data collected to study a multilayer diamond-like carbon film sample using a nondestructive testing procedure known as Neutron Depth Profiling.
Measurement Science & Technology
Bayesian, linear regularization, sensor fusion, signal processing
and Coakley, K.
Adaptive Use of Prior Information in Inverse Problems: An Application to Neutron Depth Profiling, Measurement Science & Technology
(Accessed March 2, 2024)