摘要
图像的代数特征是图像的一种固有的特征。每一幅图像都可以认为是一个矩阵,而矩阵的奇异值具有很多适合图像识别的特征。本文不仅研究了矩阵的奇异值的代数特征,而且找到一种将奇异值特征应用到目标识别中的方法,实验结果证明应用奇异值特征可以取得较好的识别效果。
Algebraic features represent the intrinsic attributions of an image. Any image can be considered as a matrix. Singular values (SV) of the matrix have some important properties adaptable to image recognition. In this paper the SV algebraic features are proved, and a method to apply the SV feature to target recognition is described. The simulation results show that the SV feature has good performance of recognition.
出处
《系统工程与电子技术》
EI
CSCD
1993年第3期14-17,共4页
Systems Engineering and Electronics
关键词
特征选择
矩阵
目标识别
模式识别
Feature extraction, Singular values feature, Circulation matrix.
