Multi-modality face recognition: An information theoretic approach

Musica Supriya, Sanjay Singh

Abstract: Face recognition has gained a great importance in recent years due to the increasing demands of the real-world applications. The modality-independent face recognition also known as heterogeneous face recognition is useful in many applications. Here modality refers to different lighting scenarios in which the picture of the subject is taken. Modality-independent face recognition addresses the issues of low illumination, discrepancies based on the modalities that is recognition of one image from the other image captured from two different sources for the same subject. To address this issue we have made use of mutual information concept. The mutual information between input and gallery images has been used to find the appropriate match between multi-modal images. The idea here is to find the appropriate match between a sketch and a photograph of the same subject. The proposed methodology is useful in crime investigation, law enforcement and for surveillance purposes.