基于谱共轭梯度法的图像极小化去噪模型算法  

Image Minimization Denoising Model Algorithm Based on Spectral Conjugate Gradient Method

作　　者：夏福全[1] 陈龙卫[2] XIA Fuquan;CHEN Longwei(School of Science, Bengbu University, Bengbu 233000, China;Taizhou Higher Vocational School of Mechanical & Electrical Technology, Taizhou 225300, China)

机构地区：[1]蚌埠学院理学院,安徽蚌埠233030 [2]泰州机电高等职业技术学校,江苏泰州225300

出　　处：《广东石油化工学院学报》2019年第3期42-46,共5页Journal of Guangdong University of Petrochemical Technology

基　　金：安徽省高校自然科学研究一般项目(113052015KJ06)

摘　　要：提出一种新的不依赖线搜索的谱共轭梯度算法,证明了新算法的全局收敛性。通过含有参数泛函的极小化去噪模型结合新的谱共轭梯度算法,从而达到恢复噪声图像的目的。数值实验结果表明,新算法在收敛速度和迭代次数等方面均有较好的数值结果,同时也能起到较好的图像去噪效果。A new spectral conjugate gradient algorithm independent of line search is proposed to prove the global convergence of the new algorithm. The minimization denoising model with parametric functionals is combined with the new spectral conjugate gradient algorithm to achieve the purpose of noise image restoration. The numerical experiments show that the new algorithm has good numerical results in terms of convergence speed and number of iterations, and also plays a good image denoising effect.

关 键 词：谱共轭梯度法 全局收敛性 图像去噪 椒盐噪声 

分 类 号：O221.2[理学—运筹学与控制论]