一种改进的视频序列超分辨率重建算法及应用  

An Improved Super-resolution Reconstruction Algorithm of Video Sequence and Its Application

作　　者：周芳[1,2] 蒋建国[1] 王培珍[2] 

机构地区：[1]合肥工业大学计算机与信息学院,安徽合肥230009 [2]安徽工业大学电气信息学院,安徽马鞍山243002

出　　处：《工程图学学报》2011年第1期45-51,共7页Journal of Engineering Graphics

基　　金：国家自然科学基金资助项目(50874001);国家教育部博士点基金资助项目(20060359004);安徽工业大学校青年教师科研基金资助项目(QZ201013)

摘　　要：在凸集优化基础上,充分利用最大后验概率和凸集投影技术,提出了一种高效强鲁棒性视频序列分辨率提升算法。首先,在空域设计一个简单的预处理共轭梯度估计器,预测原始高分辨率图像;然后,在小波域分别创建帧间和帧内两个不同的凸集,并实施不同的投影运算,提取出隐含在相邻低分辨率图像中的细节信息;最后,利用空域估计器中相邻因子间的关系约束凸集投影解的可行域,保证快速获得图像重建的唯一最优解。仿真实验和实际交通监测系统应用结果均表明,该方法较其他方法不仅可获得更高的峰值信噪比和更好的可视化效果,而且收敛更快,鲁棒性更强。An improved reconstruction method to enhance the image resolution for multi-frame video sequence is proposed based on jointly making use of the techniques of Maximum A Posteriori Probability（MAP） and Project Onto Convex Sets（POCS）.At first,a simplified spatial-domain estimator is employed by introducing the preconditioned conjugate gradient method to forecast the original high-resolution image.Then,under the wavelet-domain,two different convex sets and projection operators are designed from the aspects of inter-frame and intra-frame to extract the details hidden among the adjacent observed low-resolution frames.At last,taking advantage of the relationship between the neighborhood factors in the spatial estimator to put constraints on the potential solutions of the POCS,it can get uniqueness optimal solution quickly.The results of the simulation experiment and practical application of supervisory system demonstrate that the proposed method not only outperforms the other methods on the aspects of PSNR（Peak Signal-to-Noise Ratio） and visual quality of the reconstructed sequence,but also has the advantages of rapid convergence and great robust.

关 键 词：计算机应用 超分辨率重建 预处理共轭梯度 凸集投影 视频序列 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]