Robust Face Recognition Using A Hybrid Distance Measurement
| Autor: | Chu-Yang Wang, 王巨陽 |
|---|---|
| Rok vydání: | 2011 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 99 n the process of face recognition, a nearest neighbor classifier is usually used for classification. And the distance measures used in the nearest neighbor classifier will greatly affect the recognition performance. In this thesis, we proposed a hybrid distance measurement to measure how similar between two feature matrices that extracted form face images. In 2004, Yang et al. [10] proposed a new technique called two-dimensional principal component analysis (2DPCA). As opposed to PCA, 2DPCA is based on 2D image matrices rather than 1D vectors so the image matrix does not need to be transformed into a vector prior to feature extraction. In 2005, Zhang et al. [24] proposed another feature extraction called two-directional two-dimensional PCA ((2D)2PCA), which has more advantage than 2DPCA. But there is few people pay necessary attention to the classification measures. The typical classification measure used in 2DPCA is the sum of the Euclidean distance between two feature vectors in feature matrix. This proposed method is to compute a new distance to measure how similarity of two 2D samples. To test its performance, experiments are done based on ORL and AR face databases. The experimental results show this proposed is more efficient than the typical classification measure used in 2DPCA and more robust in different facial images conditions and face databases. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|