Raw格式岩心图像超分辨率重建  

Super-resolution reconstruction of Raw core image

作　　者：黄帅坤 陈洪刚[1] 卿粼波[1] 郝传铭 Huang Shuaikun;Chen Honggang;Qing Linbo;Hao Chuanming(College of Electronics and Information Engineering,Sichuan University,Chengdu 610065,China)

机构地区：[1]四川大学电子信息学院,四川成都610065

出　　处：《信息技术与网络安全》2020年第10期1-6,共6页Information Technology and Network Security

摘　　要：在岩心面阵相机开发中,可以使用基于学习的超分辨率技术来提升岩心图像的分辨率。针对现有超分辨率技术在重建岩心图像时存在的细节模糊或色彩偏差等问题,提出了一种基于深度卷积神经网络的Raw格式岩心图像超分辨率重建算法。首先,模拟相机图像处理器的线性处理部分合成线性图像数据集;然后,通过一个双层卷积神经网络,分别训练高低分辨率图像之间的纹理、色彩映射关系;最后,用重建出的线性高分辨率图像模拟相机图像处理器的非线性处理部分,获得纹理清晰且色彩逼真的岩心重建图像。实验结果表明,本文提出的重建算法提升了岩心图像的重建效果。In the development of core array camera,the learning based super resolution technology can be used to im-prove the resolution of core image.In order to solve the problems of detail blur or color deviation in the reconstruction of core images by existing super-resolution technologies,this paper proposes a Raw core image super-resolution recon-struction algorithm based on deep convolutional neural network.Firstly,the linear processing part of the analog camera image processor synthesizes the linear image data set.Then,a two-layer convolutional neural network is used to train the texture and color mapping relationship between high and low resolution images.Finally,the reconstructed linear high-resolution image simulates the nonlinear processing part of the camera image processor to obtain the core reconstruction image with clear texture and realistic color.Experimental results show that the reconstruction algorithm proposed in this paper improves the reconstruction effect of core images.

关 键 词：Raw格式图像 线性图像 超分辨率重建 卷积神经网络 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP183[自动化与计算机技术—计算机科学与技术]