高效全局注意网络的图像超分辨率重建  

Efficient Global Attention Networks for Image Super-Resolution Reconstruction

作　　者：王庆庆 辛月兰 赵佳 郭江 王浩臣 Wang Qingqing;Xin Yuelan;Zhao Jia;Guo Jiang;Wang Haochen(The College of Computer,Qinghai Normal University,Xining 810001,Qinghai,China;The State Key Laboratory of Tibetan Intelligent Information Processing and Application,Xining 810001,Qinghai,China)

机构地区：[1]青海师范大学计算机学院,青海西宁810001 [2]省部共建藏语智能信息处理及应用国家重点实验室,青海西宁810001

出　　处：《激光与光电子学进展》2024年第10期369-379,共11页Laser & Optoelectronics Progress

基　　金：国家自然科学基金项目(61662062);青海省自然科学基金面上项目(2022-ZJ-929)。

摘　　要：针对现有高效超分辨率重建算法大多集中于减少参数量,缺乏对层次特征的关注,存在图像高维特征不能得到充分利用的问题,提出一种高效全局注意网络。该网络的主要思想是设计交叉自适应特征块对图像进行不同层次的深度特征提取,以改善图像高频细节信息缺失的问题。另外,构造了近邻像素重构块,将空间关联性和像素分析相结合,进一步促进边缘细节信息的重建。此外,还提出一种多阶段动态余弦热重启训练策略,通过动态调整学习率以避免模型过拟合,提高训练过程的稳定性并优化网络性能。大量实验结果表明,所提方法在Set5等5个基准数据集上相比于其他先进网络,峰值信噪比(PSNR)和结构相似性(SSIM)性能指标平均提高了0.51 dB和0.0078,参数量和浮点运算量(FLOPs)平均降低了332×10^(3)和70×10^(9)。综上,所提方法在拥有较低复杂度的同时,获得了更好的性能指标和视觉效果,实现了网络高效化。To address the prevalent focus on reducing the parameter counts in current efficient super-resolution reconstruction algorithms,this study introduces an innovative efficient global attention network to solve the issues regarding neglecting hierarchical features and the underutilization of high-dimensional image features.The core concept of the network involves implementing cross-adaptive feature blocks for deep feature extraction at varying image levels to remove the insufficiency in high-frequency detail information of images.To enhance the reconstruction of edge detail information,a nearest-neighbor pixel reconstruction block was constructed by merging spatial correlation with pixel analysis to further promote the reconstruction of edge detail information.Moreover,a multistage dynamic cosine thermal restart training strategy was introduced.This strategy bolsters the stability of the training process and refines network performance through dynamic learning rate adjustments,mitigating model overfitting.Exhaustive experiments demonstrate that when the proposed method is tested against five benchmark datasets,including Set 5,it increases the peak signal-to-noise ratio(PSNR)and structural similarity(SSIM)performance metrics by an average of 0.51 dB and 0.0078,respectively,and trims the number of parameters and floating-point operations(FLOPs)by an average of332×10^(3)and70×10^(9)compared with leading networks.In conclusion,the proposed method not only reduces complexity but also excels in performance metrics and visualization,thereby attaining remarkable network efficiency.

关 键 词：图像超分辨率重建 高效全局注意 层次特征 像素重构 训练策略 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]