Subspace Framework for Feature-Level Fusion with Its Application to Handmetric Verification

A framework of FLF (feature-level fusion) is put forward based on subspace analyzes. While verifies that existing feature concatenating methods are instances of the framework, a parameter optimized model using KPCA (kernel principal components analyze) is presented and applied to the fusion of palmprint, knuckleprint and hand shape, which is called handmetric verification. The experiment testifies the effectiveness of proposed method