Kinship Verification Using Facial Images Based On Local Discrimination

Facial images convey many important human characteristics,such as identity,gender,expression,age,ethnicity and so on.Over the past two decades,a large number of face analysis problems have been investigated in the computer vision and pattern recognition community.Kinship verification from facial images is a new challenging problem in computer vision,and there are few attempts on tackling this problem.In this paper,we use the method of traditional feature extraction method,Fisher discriminant analysis and canonical correlation analysis,combined with the methods of local discrimination,extract the features of the traditional methods used in the identification of kinship.Experimental results validates that the proposed methods can gain comparatively well classification performance.The main works of the thesis are listed below:(1)the Euclidean distance between sample points is used to measure the similarity of the data samples,the high dimension of face data are projected to the low dimensional space to minimize the distance between kinship-related data samples in the projected space and maximize the distance between data samples without kinship relation,so as to retain the information with discrimination ability.The between-class scatter matrix and within-class scatter matrix are constructed,according to the characteristics of the data samples of kinship verification.We construct the balance parameter is applied in our method to better maintain the local structure and neighborhood information of data in the project process.(2)Human face kinship datasets hold small similarity within the same sample group and obvious difference between groups,we apply different models for parents and children and introduce Canonical Correlation Analysis(CCA)into kinship verification which focus on multi-modal identification which can maximize the correlation between different modal data and reduces the uncertainty of the dat a samples,so as to achieve the purpose of enhancing the ability of recognition.Based on the advantages of CCA,we use the method of local discriminative CCA which introduces the class information of samples.Finally,fusion the feature extracted from parents model and the feature extracted from children model by parallel is used to complete the recognition of kinship.