基于最小l1稀疏图表学习分类的图像识别方法研究
Study on an image recognition method based l1-minimization graphs
蒋业文 1于昕梅1
作者信息
- 1. 佛山科学技术学院电子与信息工程学院,广东佛山528000
- 折叠
摘要
利用信号的稀疏性建立图像分类处理模型是图像识别技术的新应用.通过分析最小l1范数稀疏性的原理,本文导出了一种最小l1范数稀疏性十字“花束”多面体实现模型,并在此基础上,构造了一种l1图表学习分类算法.通过与几种常用的图像分类算法比较,实验结果说明,本文提出的l1图表学习分类算法具有更高的分类精度和有效性.
Abstract
Image classification model utilizing signal sparsity is a novel application for image recognition technology.Based on l1-minimization norm sparsity,the paper exports a l1-minimization sparsity "cross-and-bouquet" polytope model,and constructs al1-graphs clustering algorithm.Comparing a few of conventional image clustering algorithm,theoretical results show thut the l1-graphs clustering algorithm possesses of much better accuracies and the effectiveness.
关键词
最小l1/稀疏性/图像分类/图表Key words
l1-minimization/sparsity/image classification/graph引用本文复制引用
基金项目
广东省科技计划项目(2009B050800004)
出版年
2013