基于小递归卷积神经网络的图像超分辨算法

马昊宇; 徐之海; 冯华君; 李奇; 陈跃庭

doi:doi:10.3788/gzxb20184704.0410004

光子学报, 2018, 47 (4): 0410004, 网络出版: 2018-03-15

基于小递归卷积神经网络的图像超分辨算法

Image Super-resolution Based on Tiny Recurrent Convolutional Neural Network

论文大纲

马昊宇 ^*徐之海冯华君李奇陈跃庭

作者单位

浙江大学光电工程与信息学院现代光学仪器国家重点实验室, 杭州 310027

超分辨成像卷积神经网络递归神经网络图像恢复底层视觉 Super-resolution Convolutional neural network Recurrent neural network Image restoration Low-level vision

摘要

针对现有软件实现超分辨算法通常过于复杂、运算开销大、模型复杂度高的问题, 本文从成像过程中图像退化的物理原理出发, 提出一套基于小递归卷积神经网络的单帧图像超分辨模型.将物理模型的约束融入到模型中, 与现有的基于统计学习的图像超分辨算法相比, 本文提出的模型的模型复杂度和计算量几乎可以忽略不计, 同时内部的参数也有着更加明确的物理意义, 并且引入了外部数据辅助对相应的模型参数进行学习.使用运行速度、峰值信噪比的数值方法对结果进行评价, 结果表明: 本文提出的算法消耗时间只有传统反向投影算法的75%, 而精度比反向投影算法提高了0.2 dB, 比双线性插值提高了1.2 dB.本文提出的算法可以取得比迭代反投影算法更快、重建精度更高的超分辨重建效果.

Abstract

A super-resolution algorithm via tiny recurrent convolutional neural network was proposed on the basis of the principles of image degradation. The proposed model has very few parameters when compared to super-resolution algorithms based on naive statistical learning. Model parameters of the proposed model have their specific physical meanings because corresponding image degradation model is introduced and regularizes the proposed model implicitly. This paper also provides an inner view of the related parameters of the algorithm and how these parameters influence the performance of the algorithm. As a result, the proposed model can achieve better performance in terms of running speed and peak signal noise ratio, comparing to current iterative backprojection algorithm. The result illustrates that the proposed algorithm only takes about 75% time consumpion, but improves the peak signal noise ratio by 0.2 dB comparing to conventional backprojection algorithm and 1.2 dB improvement comparing to bilinear interpolation respectively.

参考文献

[1] 姚保利, 雷铭, 薛彬,等. 高分辨和超分辨光学成像技术在空间和生物中的应用［J］. 光子学报, 2011, 40(11): 1607-1618.

YAO Bao-li, LEI Ming, XUE Bin, et al. Progress and applications of high-resolution and super-resolution optical imaging in space and biology［J］. Acta Photonica Sinica, 2011, 40(11): 1607-1618.

[2] 刘海英, 李云松, 吴成柯. 一种数字微镜阵列分区控制和超分辨重建的压缩感知成像法［J］. 光子学报, 2014, 43(5): 0510002.

LIU Hai-ying, LI Yun-song, WU Cheng-ke. A method for compressive sensing of images based on zone control of digital micromirror device and super-resolution［J］. Acta Photonica Sinica, 2014, 43(5): 0510002.

[3] SCHOLEFIELD A, DRAGOTTI P L. Accurate image registration using approximate Strang-Fix and an application in super-resolution［C］. Signal Processing Conference, IEEE, 2010: 1063-1067.

[4] YANG J, WRIGHT J, HUANG T S,et al. Image super-resolution via sparse representation［J］. IEEE Transactions on Image Processing, 2010, 19(11): 2861-2873.

[5] ZEYDE R, ELAD M, PROTTER M. On single image scale-up using sparse-representations［M］. Curves and Surfaces. Springer Berlin Heidelberg, 2010: 711-730.

[6] TIMOFTE R, DE V, GOOL L V. Anchored neighborhoodregression for fast example-based super-resolution［C］. IEEE International Conference on Computer Vision, IEEE, 2014: 1920-1927.

[7] TIMOFTE R, SMET V D, GOOL L V. A+: Adjusted anchored neighborhood regression for fast super-resolution［C］. Asian Conference on Computer Vision. Springer, Cham, 2014: 111-126.

[8] SCHULTER S, LEISTNER C, BISCHOF H. Fast and accurate image upscaling with super-resolution forests［C］. Computer Vision and Pattern Recognition, IEEE, 2015: 3791-3799.

[9] ROMANO Y, ISIDORO J, MILANFAR P. RAISR: rapid and accurate image super resolution［J］. IEEE Transactions on Computational Imaging, 2017, 3(1): 110-125.

[10] DONG C, CHEN C L, HE K,et al. Learning a deep convolutional network for image super-resolution［C］. European Conference on Computer Vision, Springer, Cham, 2014: 184-199.

[11] DONG C, CHEN C L, HE K,et al. Image super-resolution using deep convolutional networks［J］. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2016, 38(2): 295-307.

[12] TIMOFTE R, ROTHE R, GOOL L V. Seven ways to improve example-based single image super resolution［C］. Computer Vision and Pattern Recognition, IEEE, 2016: 1865-1873.

[13] LEDIG C, THEIS L, HUSZAR F,et al. Photo-realistic single image super-resolution using a generative adversarial network［J］. Arxiv Preprint Arxiv, 2016: 1609.04802.

[14] BASE L T. neural networks for pattern recognition［M］. Oxford University Press, 1995.

[15] HE K, Z X, REN S, et al. Deep residual learning for image recognition［C］. Computer Vision and Pattern Recognition, IEEE, 2016: 770-778.

马昊宇, 徐之海, 冯华君, 李奇, 陈跃庭. 基于小递归卷积神经网络的图像超分辨算法[J]. 光子学报, 2018, 47(4): 0410004. MA Hao-yu, XU Zhi-hai, FENG Hua-jun, LI Qi, CHEN Yue-ting. Image Super-resolution Based on Tiny Recurrent Convolutional Neural Network[J]. ACTA PHOTONICA SINICA, 2018, 47(4): 0410004.

基于小递归卷积神经网络的图像超分辨算法

关于本站 Cookie 的使用提示

全站搜索

基于小递归卷积神经网络的图像超分辨算法

相关论文

相关资讯

关于本站 Cookie 的使用提示

全站搜索