Scholarlink

<p>The complex electromagnetic environment inside the GIS equipment and limitations of the optical imaging system cause a deviation between the ideal image point and the actual projection point of the monitoring image, leading to distortion. The main reason for the distortion compensation effect is that the mathematical relationship between the ideal image point and actual projection point cannot be accurately determined. This paper proposes an optimization study of a distortion compensation algorithm for a GIS equipment compact structure monitoring images based on the LM-BP neural network algorithm. It analyzes the imaging characteristics and distortion causes in the acquisition process, builds a distortion model, and determines distortion parameters from radial and tangential perspectives. A BP neural network is used to build the distortion-compensation model, taking actual projection points as input and outputting ideal image points through sample data training. The LM optimization algorithm is used to the BP neural network's reverse error calculation process, improving convergence, reducing calculation error, and achieving accurate compensation. The experimental results show that compensated image points are highly consistent with ideal image points, effectively eliminating distortion, reducing cross entropy, and improving the peak signal to noise ratio of the monitoring image.</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

Optimization of distortion compensation algorithm for GIS equipment compact structure monitoring image based on LM-BP neural network algorithm

  • 期刊论文

  • 发表日期：

    2025年5月12日

查看全文
  • Pei Cao,
  • Peng Xu,
  • Guliang Zhou,
  • Wei Liao,
  • Kai Gao,
  • Yikai Yan,
  • Jie Yang

摘要

The complex electromagnetic environment inside the GIS equipment and limitations of the optical imaging system cause a deviation between the ideal image point and the actual projection point of the monitoring image, leading to distortion. The main reason for the distortion compensation effect is that the mathematical relationship between the ideal image point and actual projection point cannot be accurately determined. This paper proposes an optimization study of a distortion compensation algorithm for a GIS equipment compact structure monitoring images based on the LM-BP neural network algorithm. It analyzes the imaging characteristics and distortion causes in the acquisition process, builds a distortion model, and determines distortion parameters from radial and tangential perspectives. A BP neural network is used to build the distortion-compensation model, taking actual projection points as input and outputting ideal image points through sample data training. The LM optimization algorithm is used to the BP neural network's reverse error calculation process, improving convergence, reducing calculation error, and achieving accurate compensation. The experimental results show that compensated image points are highly consistent with ideal image points, effectively eliminating distortion, reducing cross entropy, and improving the peak signal to noise ratio of the monitoring image.