摘要
移动互联网的发展以及社交信息技术的不断涌现使得大量的数据信息充斥在生产和生活过程中,其中图像作为事物的直观呈现方式之一,其应用范围得到较大程度的扩展。但图像融合质量、成像噪声、像素分辨率等因素使得当前计算机图像重建质量难以得到较好的保证。基于凸集投影算法,考虑到迭代次数和松弛因子对算法的影响,对其进行全变分处理以保证图像的边缘信息。同时为减少算法步长对收敛性能的干扰,在全变分凸集投影算法基础上引入Armijo法则和KSVD分解以实现目标图像训练和稀疏表达。结果表明,所提算法MSE均值为355.28,平均SNR为21.44,信息熵值基本低于8.5,其性能结果均优于其他四种对比算法,图像重建信息质量效果较好,能有效为计算机图像信息的处理和伪影消除提供新的改进思路。
The development of mobile Internet and the continuous emergence of social information technology make a large amount of data information flooded in the process of production and life,in which image,as one of the intuitive presentation methods of things,has been greatly expanded in its application scope.However,image fusion quality,image noise,pixel resolution and other factors make it difficult to ensure the quality of current computer image reconstruction.Therefore,the research is based on convex set projection algorithm.Considering the influence of iteration number and relaxation factor on the algorithm,the full variation processing is carried out to ensure the edge information of the image.At the same time,in order to reduce the interference of algorithm step size on the convergence performance,Armijo rule and KSVD decomposition are introduced on the basis of the total variation convex set projection algorithm to achieve target image training and sparse expression.The results show that the MSE average value of the proposed algorithm is 355.28,the average SNR is 21.44,and the information entropy value is basically lower than 8.5.Its performance results are better than the other four comparison algorithms.The image reconstruction information quality is good,which can effectively provide a new idea for the computer image information processing and artifact elimination.
作者
王振华
WANG Zhen-hua(Intelligent Industry College of Fuzhou Software Vocational and Technical College,Fuzhou 350001,Fujian,China)
出处
《贵阳学院学报（自然科学版）》
2023年第3期89-94,共6页
Journal of Guiyang University：Natural Sciences
