稳健峰值感知引导滤波并应用于信号保峰平滑

ue to environmental and equipment constraints, low signal-to-noise ratio is frequently observed in collected signals, significantly affecting the quality of the signal and the accuracy of subsequent analysis. Traditional signal smoothing methods often result in over-smoothing while removing noise. In order to eliminate noise while preserving peak information, this paper proposes a robust peak-aware guided filtering model. By designing robust peak-aware weights based on local signal variance and hyperbolic tangent functions, a robust peak-aware guided filtering model (RPAGF) is introduced by combining these weights with guided filtering. The overlapping window processing in the filtering output utilizes a weighted averaging step to enhance the anisotropy of the filtering model, with the guided signal selection being the filtered output of a time fractional-order supRobust peak-aware guided filter for signal peak-preservinger-diffusion model. Experimental results demonstrate the outstanding peak-preserving smoothing performance of RPAGF.