摘要
Nearest feature line (NFL) is a powerful tool in pattern recognition, which generalises the representational capacity of available prototypes by using linear interpolation and extrapolation between the feature points. NFL is generalised to subspace learning so that the obtained subspace has desirable discriminating ability. Experiments on face recognition demonstrate its effectiveness.
源语言 | 英语 |
---|---|
页(从-至) | 1079-1080 |
页数 | 2 |
期刊 | Electronics Letters |
卷 | 43 |
期 | 20 |
DOI | |
出版状态 | 已出版 - 2007 |
已对外发布 | 是 |