|
|
LLE-Based Classification Algorithm and Its Application to Passive Millimeter Wave Target Recognition |
Luo Lei; Li Yue-hua |
School of Electronic Engineering and Optoelectronic Technology, Nanjing University of Science and Technology, Nanjing 210094, China |
|
|
Abstract A new one-class classification algorithm is put forward based on the idea of Locally Linear Embedding (LLE) and the low dimensional manifold of data distribution, for the issue of one-class classification of pattern recognition. The low dimensional manifold the short-time Fourier spectrum of passive millimeter wave (MMW) detector echo signal is found, which characteristics are explored as well. The algorithm is more efficient than present popular one-class classification algorithm when applied to passive MMW metal target recognition. Moreover, it is not sensitive to input parameters and is robust to data aliasing.
|
Received: 12 June 2009
|
|
Corresponding Authors:
Li Yue-hua
|
|
|
|
|
|
|