%0 Journal Article
%T 基于信息熵和OIF算法的数据降维<br>Data Dimension Reduction Based on Information Entropy and OIF Algorithms
%A 吴逍航
%A 付小宁
%A 秦帅
%J Computer Science and Application
%P 1365-1376
%@ 2161-881X
%D 2019
%I Hans Publishing
%R 10.12677/CSA.2019.97154
%X <div style=\"text-align:justify;\">
	<span style=\"font-size:14px;\">自动子空间划分(ASP)和最佳指数法(OIF)广泛应用于高光谱数据降维。在ASP和OIF算法的基础上引入信息熵和巴氏系数，提出BSEF算法。该方法首先通过ASP对高光谱波段进行波段划分；以信息熵衡量波段信息量，以巴氏系数衡量波段之间的相关性，选出信息量大且相关性小的波段；在BSEF算法基础上，提出BERF数据降维算法运用于有监督目标检测。实验证明：BSEF可降低算法运行时间至原来的四分之一，BERF可降低算法运行时间至原来的三分之一，两种降维方法均可以保证算法的检测效率。<br />
Automatic subspace partitioning (ASP) and optimum index factor (OIF) are widely used in dimensionality reduction of hyperspectral data. On the basis of ASP and OIF algorithm, information entropy and Barkhault coefficient are introduced, and BSEF algorithm is proposed. Firstly, it divides the hyperspectral bands by ASP. The band information is measured by information entropy, and the correlation between bands is measured by Barkhault coefficient. The bands with large infor-mation and small correlation are selected. The BERF dimensionality reduction algorithm for supervised target detection is proposed based on BSEF algorithm. Experiments show that BSEF can reduce the running time of the algorithm to a quarter of the original time, and BERF can reduce the running time to a third of the original time. Both dimensionality reduction methods can ensure the detection efficiency.</span> 
</div>
%K 信息熵，巴氏系数，OIF算法，数据降维<br>Information Entropy
%K Barkhausen Coefficient
%K OIF Algorithm
%K Data Dimensionality Reduction
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=31404