图像分类的字典学习方法概述
1 字典学习(dictionary learning)
旨在从原始数据中找到一组特殊的稀疏信号,在机器视觉中称为视觉单词(visual words),这一组稀疏元素能够足够线性表示所有的原始信号。字典学习来源于压缩感知,后来广泛用于图像去噪、去雾、聚类、分类等方面。
两类:
1) 直接学习区分性的字典(directly forcing the dictionary discriminative)Track I
2)稀疏化系数,使得到的子弟那具有可区分性(making the sparse coefficients discriminative (usuallythrough simultaneously learning a classifier) to promote the discrimination ofthe dictionary.)Track II
2 字典学习的原理
假设为原始数据,字典学习的目的是通过下面的优化得到可区分性能很好表示原始数据的字典:
(1) Track I: Directly Making theDictionary Discriminative
The methods from Track I use the reconstruction error for the finalclassification, thus the learned dictionary ought to be as discriminative aspossible.
1)Meta-face learning
注:SRC(Sparse Representation-Based Classification)最开始用于鲁棒性人脸识别,C代表不同人脸个数,