Scholarlink

<p>The synchrosqueezing transform (SST) provides a sharp time–frequency (TF) representation for signals with slowly varying instantaneous frequencies (IFs). However, its performance deteriorates significantly in the presence of noise and fast IF variations, which violate the underlying assumptions of the SST. To address this limitation, this paper proposes reassigned range (RR) based SST designed to achieve noise-robust and high-resolution TF analysis. The core of the method is a&#xa0;reassigned range (RR)&#xa0;criterion, constructed through two key innovations: an&#xa0;iterative scheme&#xa0;that refines the IF estimation by progressively reducing errors induced by fast frequency modulation, and the exploitation of the&#xa0;local maximum property&#xa0;of IFs in TF plane to enhance discrimination between signal components and noise. Based on the RR, an adaptive threshold is derived to separate signal energy from noise, leading to effective noise suppression. Consequently, proposed method simultaneously improves the readability of the TF representation and attenuates noise interference. Evaluations on both synthetic non-stationary signals and real-world data demonstrate the superior performance of the proposed method compared to existing techniques.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Iterative Scheme and Local Maximum Property Based Reassigned Range used in Synchrosqueezing Transforms

  • 期刊论文

  • 发表日期：

    2026年6月23日

查看全文
  • Yang Zhou,
  • Bingo Wing-Kuen Ling

摘要

The synchrosqueezing transform (SST) provides a sharp time–frequency (TF) representation for signals with slowly varying instantaneous frequencies (IFs). However, its performance deteriorates significantly in the presence of noise and fast IF variations, which violate the underlying assumptions of the SST. To address this limitation, this paper proposes reassigned range (RR) based SST designed to achieve noise-robust and high-resolution TF analysis. The core of the method is a reassigned range (RR) criterion, constructed through two key innovations: an iterative scheme that refines the IF estimation by progressively reducing errors induced by fast frequency modulation, and the exploitation of the local maximum property of IFs in TF plane to enhance discrimination between signal components and noise. Based on the RR, an adaptive threshold is derived to separate signal energy from noise, leading to effective noise suppression. Consequently, proposed method simultaneously improves the readability of the TF representation and attenuates noise interference. Evaluations on both synthetic non-stationary signals and real-world data demonstrate the superior performance of the proposed method compared to existing techniques.