低秩判别性字典学习及组织病理图像分类算法  

Discriminative Dictionary Learning with Low-rank Constraint for Histopathological Images Classification

作　　者：毛丽珍[1] 汤红忠[1,2] 范朝冬 曾淑英[1] MAO Li-zhen;TANG Hong-zhong;FAN Chao-dong;ZENG Shu-ying(College of Information Engineering,Xiangtan University,Xiangtan 411105,China;Key Laboratory of Intelligent Computing & Information Processing of Ministry of Education,Xiangtan 411105,China)

机构地区：[1]湘潭大学信息工程学院,湖南湘潭411105 [2]湘潭大学智能计算与信息处理教育部重点实验室,湖南湘潭411105

出　　处：《小型微型计算机系统》2019年第9期1881-1885,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61573299,61602397)资助;湖南省自然科学基金项目(2017JJ33115,2017JJ2251,2016JJ3125)资助

摘　　要：针对组织病理图像分类中样本特征之间具有高度相关性的问题,本文提出了一种基于低秩约束的判别性字典学习算法,并将其应用于组织病理图像分类.与传统算法仅仅关注稀疏编码的低秩性不同,本文算法不仅同时优化了子字典对同类和非同类训练样本的重构性能,而且对类独有的子字典增加了低秩性约束.这一策略可以降低类独有的子字典原子之间的相似性,促进原子之间相互独立,从而学习出更具判别性、结构更紧凑的字典.在ADL数据集上的实验结果表明,与现有算法相比,本文提出的算法可获得更高的分类精度.Aiming at the problem that there is a high correlation among sample features in histopathological image classification,a discriminative dictionary learning algorithm with low-rank constraint( LRCDDL) is proposed and applied to histopathological image classification. Unlike the traditional algorithms which only focus on the lowrank of sparse coding,LRCDDL not only optimizes the reconstruction performance of sub-dictionary for the same and different class training samples. Moreover,lowrank constraints are added to the class-specific sub-dictionary. This strategy can reduce the correlation among the class-specific sub-dictionary atoms and promote the independence of atoms,so as to learn a discriminative and compact dictionary. The experimental results on ADL dataset show that LRCDDL can achieve higher classification accuracy than the existing algorithms.

关 键 词：低秩约束 子字典学习 判别性字典 组织病理图像 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]