UPRE方法在图像恢复正则化参数自适应选择中的应用  

Application of UPRE Method in Adaptive Selection of Regularization Parameters in Image Restoration

作　　者：加春燕[1] JIA Chunyan(School of Fundamental Education,Beijing Polytechnic College,Beijing 100042,China)

机构地区：[1]北京工业职业技术学院基础教育学院,北京100042

出　　处：《北京工业职业技术学院学报》2024年第1期23-26,共4页Journal of Beijing Polytechnic College

基　　金：2023年北京工业职业技术学院重点科研课题(BGY2023KY-47Z)。

摘　　要：大多数图像恢复都是不适定问题,需要利用正则化方法将病态方程转化为适定方程。正则化参数主要用于调节图像保真度与图像光滑度之间的平衡,与图像恢复质量的好坏有着密切关系。无偏预计风险估计(UPRE)方法可以自适应选择最佳正则化参数,基于数学理论分析和图像恢复实验,验证了该方法在图像恢复中的可行性和有效性。Most image restoration is an ill posed problem that requires the use of regularization methods to transform ill posed equations into well posed equations.The regularization parameters are mainly used to adjust the balance between image fidelity and smoothness,which is closely related to the quality of image restoration.The Unbiased Predictive Risk Estimation(UPRE)method can adaptively select the optimal regularization parameters.Based on mathematical theory analysis and image restoration experiments,the feasibility and effectiveness of this method in image restoration have been verified.

关 键 词：图像恢复 正则化参数 自适应选择 无偏预计风险估计方法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]