Performance analysis of fusion methods for EAR biometrics
P Panchakshari, Sarika Tale · 2016
Recognizing the people by their ear shape is upcoming application in the field of biometrics. Ear shape of a person does not change over his life span it is also unique and ear appearance cannot be modified by human efforts. Acquiring ear image not need an individual attention. Considering all these unique characteristics, significance in ear authentication modules has developed extremely over a past decade. The prime aim in this program is designing an authentication module using ear biometrics. The authentication framework comprises of image capture, feature extraction and classification. The physiological trait will be taken from existing benchmark databases. Global and local features will be extracted. Outer helix is considered as global feature and concha portion is used as local feature. The physiological trait will be taken from existing benchmark databases. Global and local features will be extracted. Features will be extracted using PCA and HOG. Fusion of local and global features will be analyzed. Fusion by feature level and score level matching fusion in ear biometrics will be performed. Euclidean distance will be used as a classifier.