Skip to main navigation Skip to search Skip to main content

面向非线性过零信号的加权 RANSAC 高精度零点识别技术

Translated title of the contribution: Weighted RANSAC for High-precision Zero-point identification in nonlinear Zero-crossing signals
  • Xiaoxi Wei
  • , Yun Wang*
  • , Hongwu Ru
  • , Longkang Liu
  • *Corresponding author for this work
  • Beijing Institute of Technology

Research output: Contribution to journal › Article › peer-review

Abstract

To address the issue that nonlinear zero-crossing signals are susceptible to outlier interference during zero-point identification, a high-precision zero-point identification technique based on weighted Random Sample Consensus (RANSAC)is proposed. This technique constructs an arctangent nonlinear model and introduces a Gaussian kernel weight function based on signal amplitude to achieve differentiated weighting of samples. By integrating this criterion into the iterative optimization and parameter estimation of the RANSAC algorithm, interference from edge distortion and outliers on the fitting trajectory is effectively suppressed, significantly improving zero-point identification accuracy under complex working conditions. Experimental validation was conducted based on a differential alignment dynamic angle measurement system. The results demonstrate that compared with the least squares method, the proposed method improves single measurement accuracy by approximately 52% and repeat measurement accuracy by approximately 58%. This technique provides an effective approach for high-precision zero-point identification of nonlinear zero-crossing signals and parameter estimation in scenarios such as optical precision angle measurement and angle metrology calibration.

Translated title of the contributionWeighted RANSAC for High-precision Zero-point identification in nonlinear Zero-crossing signals
Original languageChinese (Traditional)
Pages (from-to)547-554
Number of pages8
JournalGuangxue Jishu/Optical Technique
Volume52
Issue number5
DOIs
Publication statusPublished - Sept 2026
Externally publishedYes

Fingerprint

Dive into the research topics of 'Weighted RANSAC for High-precision Zero-point identification in nonlinear Zero-crossing signals'. Together they form a unique fingerprint.

Cite this