基于局部约束编码的稀疏保持投影降维识别方法研究

发布时间：2018-03-10 07:38

本文选题：目标识别　切入点：维数约简　出处：《电子学报》2016年03期 　论文类型：期刊论文

【摘要】：稀疏表示技术的引入可有效解决降维处理对图参数的依赖,但这类降维方法不能同时兼顾稀疏重构和样本数据的邻近性问题.针对该问题,本文提出了一种基于局部约束编码的稀疏保持投影降维识别方法.通过稀疏表示分类模型构建了图边权矩阵,引入局部约束因子设计了降维投影模型,推导降维求解过程,分析了本文方法与SPP(Sparse Preserving Projections)和SLPP(Soft Locality Preserving Projections)方法之间的共性和区别,最后给出了识别算法流程.采用人脸图像数据集和高分辨SAR(Synthetic Aperture Radar)图像数据集对算法的有效性进行仿真验证,由于考虑了数据间的邻近性,本文方法较传统方法可获得更好的识别性能.
[Abstract]:The introduction of sparse representation technology can effectively solve the dependence of graph parameters on dimensionality reduction processing, but this kind of dimensionality reduction method can not take into account the problem of sparse reconstruction and the proximity of sample data. In this paper, a sparse preserving projection dimensionality reduction recognition method based on local constraint coding is proposed. The graph edge weight matrix is constructed by sparse representation classification model, the reduced dimension projection model is designed by introducing the local constraint factor, and the dimensionality reduction solution process is deduced. The generality and difference between this method and SPP(Sparse Preserving projects and SLPP(Soft Locality Preserving projects are analyzed. Finally, the recognition algorithm flow is given. The validity of the algorithm is verified by using face image data set and high resolution SAR(Synthetic Aperture Radar-based image data set. Considering the proximity of data, this method can obtain better recognition performance than the traditional method.
【作者单位】：海军航空工程学院电子信息工程系;海军航空工程学院7系;海军航空工程学院信息融合研究所;
【基金】：国家自然科学基金(No.61102167;No.61302008;No.61179016;No.61102165)
【分类号】：TP391.41

【相似文献】