Improved Maximum Scatter Difference Discriminant Analysis for Face Recognition
An improved maximum scatter difference (MSD) criterion is proposed in this paper. Existing MSD model only highlights the role that within-class scatter matrix plays, while pays little attention to the action of betweenclass scatter matrix. Another weakness is that it is difficult to select an appropriate weight for within-class scatter matrix in that the range of weight is too large. In order to make MSD more suitable for classification, different weights are assigned to both between-class and within-class scatter matrix, respectively. This scheme is more convenient for operation than original MSD because it confine the range of parameter to a small area. The results of experiments conducted on AR and FERET face database indicate the effectiveness of the proposed approach.
Face recognition Feature extraction Maximum scatter difference (MSD) Weighted maximum scatter difference (WMSD)
Guan Shiliang Li Xiaodong
School of Information,Linyi Normal University, Linyi,276005,China Key Laboratory of Measurement and Control of Complex Systems of Engineering, Ministry of Education,
国际会议
2009 International Workshop on Information Security and Application(2009 信息安全与应用国际研讨会)
青岛
英文
368-371
2009-11-21（万方平台首次上网日期，不代表论文的发表时间）