Feature level fusion using palmprint and finger geometry based on Canonical Correlation Analysis

Canonical Correlation Analysis (CCA) is a standard tool in statistical analysis that measures the linear relationship between two data sets. In this paper, we present a multibiometric approach which combines palmprint feature and finger geometry feature based on Canonical Correlation Analysis. First, we use linear discriminant analysis (LDA) to build the palmprint feature. Second, the geometry feature of the middle finger is extracted. Then, these two features are fused by CCA to form a combined feature which is applied to denote the identity of a person. This method makes it possible to fuse these features mentioned above together and decrease the dimension of the fusion feature. The results of experiments conducted on a database of 86 hands (10 impressions per hand) show that the CCA-based feature level fusion method has good performance.