关键词: 奇异点, 脉冲型奇异点, 分类, Bayes决策, 分界面

Abstract: This paper proposed a new method for detecting and locating the singularities in the signal. By analyzing the characteristics of singular signal, the detection of the pulse singularities was first modeled as a classification task of two classes: one class consisted of the pulse singularities and the other contained the other points in the signal. Then, based on the Bayes decision rule and Neyman-Pearson criterion, a decision surface was derived by constraining the probability of missed detection for the class containing the pulse singularities to be fixed. As a result, a Bayes Decision Based Pulse Singularity Detection (BDPSD) method was directly developed. The experimental results on a number of artificial and real signals show that the BDPSD method can greatly improve the detection quality and locating accuracy of the pulse singularity, compared with the singularity detection method based on the wavelet transform local modulus maximum theory. This also shows that BDPSD is indeed an effective and practical singularity detection method.

Key words: singularity, pulse singularity, classification, Bayes decision, decision surface