基于图像质量加权的D-S证据理论多生物特征融合识别

Multi-biometric feature fusion identification based on D-S evident theory of image quality with different weights

对人脸和指静脉在决策层的融合识别,给出一种新的组合图像质量评价方法,并采用一种改进的基于图像质量加权的D-S证据理论融合这两种生物特征。首先,组合图像质量评价法采用清晰度和对比度、相关系数的组合方法来评价生物特征图像质量;其次,应用改进的基于图像质量的匹配值增强方法和D-S证据理论,减少了极大值图像质量指标的影响,使图像质量加权与实际情况更趋一致。实验结果表明,与没有考虑图像质量因素相比,考虑了图像质量信息的D-S证据理论融合方法提高了识别效果。

For the fusing recognition with face and finger-vein at decision level,a new quality score of image by combining three indexes was presented,and an improved fusion strategy based on D-S evident theory was adopted to fuse two biometric characteristics.At first,the quality score of image was computed by combining index of distinct,contrast and coefficient.Then an improved method based on image quality and D-S evident theory was adopted.This improved method reduced the impact of maximum of image quality score and made the weighted parameter with the actual situation more consistent.Compared with the result of D-S evident theory with no regard to image quality,the results reveal that the fusion method in this paper based on D-S evident theory taking account of image quality information improves the performance.