Dimensionality Reduction Method Based on NMF and LPP
Deqin Yan · Jilin Normal University Journal · 2011
NMF is a commonly used dimensionality reduction method in recent years which is also widely used in image retrieval,face recognition and signal processing.The weight's non-negative demand produced by decomposition made the data processing a good result.NMF in the process of decomposition without considering the geometric properties and local structure internal data,so there could be some problems in processing the data accurately.This paper puts an innovative approach which combines NMF and LPP together into image retrieval use in that LPP retains the local structure and geometric properties of data.Not only decreases negative effect in image retrieval,this method could also make signal processing more efficient.A verification based on corel database has proved that the method can indeed improve the retrieval accuracy.