Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning
===============================
利用判别特征导向的词典库学习来进行病理学图像分类
文章发表于 IEEE TRANSACTIONS ON MEDICAL IMAGING, VOL. 35, NO. 3, MARCH 2016
总体思路：提取特征 –> 对特征进行分类
本文提出了一种提取特征的算法DFDL（ Discriminative
Feature-oriented Dictionary Learning），用于学习某一类别特定的词典库，用于图片分类或者病理学图像评分。本实验在3个实验集（乳腺癌，肾肺脾ADL，脑肿瘤TCGA）上做测试。
主要算法思想，每一类生成一个词典，词典的子空间可以生成该类的特征表示；任何其他类别的表示都跟这个表示距离很大。类似于类内间距最小，类间间距最大的KNN聚类方法。
方法流程：
论文笔记：Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary LearninHistopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning 
论文笔记：Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary LearninHistopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning 论文笔记：Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary LearninHistopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning Y 是某一类，Y拔是互补类， D 是字典，s是 i 类的稀缺表示，yl 是真是label， 因此生成该式：
论文笔记：Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary LearninHistopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning 利用多次迭代求解。
主要步骤如下：
论文笔记：Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary LearninHistopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning 论文笔记：Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary LearninHistopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning 学习到的词典：
论文笔记：Histopathological Image Classification Using Discriminative Feature-Oriented Dictionary LearninHistopathological Image Classification Using Discriminative Feature-Oriented Dictionary Learning 
