Applying Bayesian decision theory to peak detection of stochastic signals

Abstract

Peak detection is a general problem in a wide range of applications. In many advanced signal processing systems, peak detection is used as a pre-processing step, and hence its accuracy for validation of output is crucial. The problem of peak detection can be divided into two stages, peak detection and peak selection and validation. Peak detection can be used in finding peaks in signals and extracting them. The second stage is peak validation in which only those peaks that are representing a special feature or event in signal should be chosen. This paper investigates peak selection and validation problems. A novel peak selection algorithm based on Bayesian decision theory is proposed. It is implemented in Matlab and experimental results show that the proposed peak detection algorithm can detect and select peaks reliably.