基于偏微分方程和机器学习的图像去噪算法  

Image Denoising Algorithm Based on Partial Differential Equation and Machine Learning

作　　者：杜渺勇 施垚 周浩 韩丹夫 DU Miaoyong;SHI Yao;ZHOU Hao;HAN Danfu(School of Science,Hangzhou Normal University,Hangzhou 311121,China)

机构地区：[1]杭州师范大学理学院,浙江杭州311121

出　　处：《杭州师范大学学报（自然科学版）》2020年第2期215-219,共5页Journal of Hangzhou Normal University(Natural Science Edition)

摘　　要：本文基于偏微分方程和机器学习相结合的方法,通过减少函数微分不变量的个数,提出新的模型来提高机器学习的效率.仿真实验显示,新的模型在程序运行时间上有了巨大的缩减,并且视觉效果与原先模型相差不大,峰值信噪比与原模型相差不大.Basing on the combination of machine learning and partial differential equation,the efficiency of machine learning is improved by reducing the number of functional differential invariants.The simulation results show that the new model greatly reduces the operation time,the visual effect and PSNR are not much different from the original one.

关 键 词：图像去噪 偏微分方程 机器学习 微分不变量 坐标下降法 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术] R50[自动化与计算机技术—计算机科学与技术]