摘要
图像是最为普及的信息载体,改善图像质量,提高图像清晰度对获得更多图像信息意义重大。本文针对深度学习在图像超分辨中的应用进行研究,指出图像超分辨采用主观评价和客观评价相结合的方法,卷积神经网络的图像超分辨算法步骤。在此基础上,对卷积神经网络算法、SC算法、Bicubic算法对图像的超分辨处理效果进行对比。结果表明,卷积神经网络算法对低分辨率图像的超分辨处理保真度高、处理精度高、相似度高,图像更加平滑。本论文的研究对图像的超分辨处理具有一定的现实参考价值。
Image is the most popular information carrier.Improving image quality and clarity is of great significance to obtain more image information.This paper studies the application of depth learning in image super-resolution,and points out that image super-resolution adopts the method of combining subjective evaluation and objective evaluation,and the image super-resolution algorithm steps of convolution neural network.On this basis,the convolution neural network algorithm,SC algorithm and Bicubic algorithm are compared in the image super-resolution processing effect.The results show that the convolution neural network algorithm has the advantages of high fidelity,high precision,high similarity and smoother image.The research of this paper has a certain practical reference value for image super-resolution processing.
作者
韩鸣
Han Ming(Shaanxi Preschool Teachers College,Xi'an 710100,Shanxi)
出处
《现代科学仪器》
2020年第3期169-172,共4页
Modern Scientific Instruments
关键词
深度学习
卷积神经网络
图像超分辨
deep learning
convolution neural network
image super-resolution
