摘要
为了提高含有椒盐噪声的图像分类的准确性,运用中值滤波法研究了含有椒盐噪声的图像去噪问题。将中值滤波法用于卷积神经网络的预处理环节,通过卷积神经网络模型对滤波后的图像进行分类,得到了以下结果:卷积神经网络模型在测试集上的准确率达到82.86%,比传统的BP神经网络模型的准确率提高19.05个百分点。
In order to improve the classification accuracy of images with salt-and-pepper noise,the denoising problem of images with salt-and-pepper noise is studied by using the median filter method.The median filter method is used in the preprocessing of convolutional neural network,and the filtered images are classified by the convolution neural network model.The following results are obtained:the accuracy of convolution neural network model on the test set is 82.86%,which is 19.05%higher than that of traditional BP neural network model.
作者
赵恩铭
杨松
姚志强
骆成军
杨燕婷
刘光宇
周豹
王辰
ZHAO Enming;YANG Song;YAO Zhiqiang;LUO Chengjun;YANG Yanting;LIU Guangyu;ZHOU Bao;WANG Chen(College of engineering,Dali University,Dali 671003,China)
出处
《新乡学院学报》
2022年第3期27-31,共5页
Journal of Xinxiang University
基金
国家自然科学基金项目(62065001,61761048)
云南省专业学位研究生教学案例库建设项目
云南省地方本科高校基础研究联合专项资金项目(2019FH001-066)
云南省科技厅青年项目(2018FD070)。
关键词
图像去噪
卷积神经网络
中值滤波技术
图像分类
image denoising
convolutional neural network
median filtering technology
image classification
