摘要
稀疏表示模型常利用训练样本学习过完备字典,旨在获得信号的冗余稀疏表示.设计简单、高效、通用性强的字典学习算法是目前的主要研究方向之一,也是信息领域的研究热点.基于综合稀疏模型的字典学习方法已经广泛应用于图像分类、图像去噪、图像超分辨率和压缩成像等领域.近些年来,解析稀疏模型、盲字典模型和信息复杂度模型等新模型的出现丰富了字典学习理论,使得更广泛类型的信号能够被"简单性"描述.本文详细介绍了综合字典、解析字典、盲字典和基于信息复杂度字典学习的基本模型及其算法,阐述了字典学习的典型应用,指出了字典学习的进一步研究方向.
The sparse model often utilizes training samples to learn an over-complete dictionary, in order to obtain the redundant and sparse representation of signals. Designing simple, effective and flexible dictionary learning algorithms is one of the main and hot research topics in the information field. The dictionary learning methods based on synthesis sparse model have been applied into image classification, image denoising, image super-resolution and compressive imaging. In recent years, analysis sparse model, blind dictionary model and information complexity model have been proposed, which enrich the dictionary learning theory in order and lead to a "simple" description for a wide range of signals. In this paper, the fundamental models and dictionary learning algorithms are introduced in detail in terms of synthesis dictionary, analysis dictionary, blind dictionary and dictionary learning based on information complexity. Typical applications of dictionary learning methods are further illustrated. Finally, the directions for further research of the dictionary learning are pointed out.
出处
《自动化学报》
EI
CSCD
北大核心
2015年第2期240-260,共21页
Acta Automatica Sinica
基金
国家自然科学基金(61471313)
河北省自然科学基金(F2014203076)资助~~
关键词
字典学习
稀疏表示
综合模型
解析模型
Dictionary learning, sparse representation, synthesis model, analysis model