Integrating a cosmetic detection scheme into face–iris multimodal biometric systems

Multimodal biometric systems generally combine information coming from more than one biometric trait to increase the accuracy of recognition systems. This paper investigates the effect of cosmetics on different fusion levels of face and iris biometrics as two well-known biometric traits. In fact, the present work proposes to involve the implementation of a common cosmetic detection face-iris system first and then fusion is performed according to the kind of cosmetic using an optimized weighting system to evaluate the performance of multimodal system under different fusion levels. Therefore, a face–iris cosmetic detection system to combine shape, texture and color information of individuals using Log-Gabor (L-Gabor) transform, Uniform Local Binary Pattern (ULBP) and Color Moments (CM) extractors is integrated into the proposed multimodal recognition system. The fusion of both modalities is performed according to the result of detection scheme using an optimized weighting system to evaluate the accuracy of the proposed multimodal method at score and feature level fusions. Prior to extracting the features and fusion of modalities the present work proposes an image-edge smoothing scheme to improve the recognition accuracy of the face-iris multimodal system. The experimental results demonstrate the overall performance improvement of the proposed multimodal system under cosmetics.