基于非下降线搜索的改进PRP共轭梯度方法及在图像恢复中的应用
Improved PRP Conjugate Gradient Method Based on Non-descent Line Search and Application on Image Restoration
李朋原1
作者信息
- 1. 郑州警察学院 图像与网络侦查系,河南 郑州 450000
- 折叠
摘要
PRP方法是最有效的非线性共轭梯度优化方法之一,然而该方法不能保证产生目标函数的下降方向,这给一般函数的全局收敛带来了困难.为了保证PRP方法的全局收敛性,提出了一种改进的PRP共轭梯度方法.文章以非凸优化问题为目标,简要介绍了非下降线搜索技术以及一些适当的假设条件,探讨了改进PRP方法的全局收敛性.基于MATLAB软件工具,验证了新方法在处理无约束优化和图像恢复问题时的有效性和实用性.
Abstract
PRP method is one of the most effective methods for nonlinear Conjugate Gradient optimization.However,this method cannot guarantee the decreasing direction of the objective function,which makes the global convergence of the general function difficult.In order to ensure the global convergence of PRP method,an improved PRP Conjugate Gradient Method is proposed.Aiming at the non-convex optimization problem,this paper briefly introduces the non-descent line search technique and some appropriate assumed conditions,and discusses the global convergence of the improved PRP method.Based on MATLAB software tool,the effectiveness and practicability of the new method for processing the unconstrained optimization and image restoration problems are verified.
关键词
共轭梯度方法/非下降线搜索/全局收敛性/无约束优化/图像修复Key words
Conjugate Gradient Method/non-descent line search/global convergence/unconstrained optimization/image restoration引用本文复制引用
出版年
2024
国家科技期刊平台
NSTL
万方数据