Aerial Image Information Extraction Based on Non-negative Matrix Factorization

• 作 者: Hao Hong Xu Changqing Zhang Xinping
• 机 构: School of Environmental and Resource Sciences,Zhejiang A & F University,
• 出 处: Chinese Forestry Science and Technology
• 发表日期: 2012年03期
• 关键词: factorization superiority aerial nonnegative columns

This study was on superiority of the non- negative matrix factorization(NMF) algorithm for application of information extracted with aerial images.First,NMF was used for aerial image information extraction,and then this data was compared with a principal component analysis(PCA) in which r(the number of rows or columns of basic matrix) and E_(ignum)(the number of eigenvalues) were given different values.Experimental results showed that the run time of NMF with r = 20 or 50 was less than that of PCA with an E_(ignum) = 20 or 50.Also,the recognition rate of NMF with r = 50 was higher than that of an E_(ignum) = 50.The experiment showed that nonnegative matrix factorization had advantages of a short time period with a high recognition rate.

This study was on superiority of the non-negative matrix factorization (NMF) algorithm for application of information extracted with aerial images. First, NMF was used for aerial image information extraction, and then this data was compared with a principal component analysis (PCA) in which r (the number of rows or columns of basic matrix) and E_ (ignum) (the number of eigenvalues) were given different values. Experimental results showed that the run time of NMF with r = 20 or 50 was less than that of PCA with an E - (ignum) = 20 or 50. Also, the recognition rate of NMF with r = 50 was higher than that of an ignores = 50. The experiment showed that nonnegative matrix factorization had advantages of a short time period with a high recognition rate.