稀疏表示在脑部图像融合研究中的进展

Progress of sparse representation in image fusion

对稀疏表示理论在脑部图像融合的应用进行了综述。围绕其算法流程对相关原理及研究现状分别进行阐述。在字典构造方面,围绕K奇异值算法框架对其展开讨论;在稀疏重构方面,就正交匹配追踪算法进行简要描述;在稀疏活跃度水平计算方面,以一种加权多范数活跃度度量方法为例进行分析。此外,概述了稀疏表示在脑部图像融合中的处理规则,讨论其面临的挑战,对相关工作的研究提出了期待。完整地梳理了稀疏表示理论的原理及其在脑部图像融合应用中的基本规则,可对该领域的工程实践提供理论研究基础。

The application of sparse representation theory in brain image fusion is reviewed, and the relevant principles and research status are described respectively around its algorithm process. In terms of dictionary construction, the algorithm framework of K-singular Value Decomposition(K-SVD) is discussed. In the aspect of sparse reconstruction, Orthogonal Matching Pursuit(OMP) algorithm is briefly described. In terms of sparse activity level calculation, a weighted multi-norm activity measurement method is taken as an example. In addition, the processing rules of sparse representation in brain image fusion are summarized, its challenges are discussed, and the related research is expected. In this paper, the principle of sparse representation theory and the basic rules of its application in brain image fusion are thoroughly reviewed, which is of guiding significance to the theoretical research and engineering practice in this field.