The buaa-visnir face database instructions

As the problem of face recognition has been studied for over 30 years, traditional 2D images have been proved inadequate for some challenging problems such as surveillance at night and face recognition out door because the variations of pose and illumination are two crucial factors which limited the rate of accuracy. Recently, more and more researchers in face recognition area fall to work dealing with new types of data like 3D images and heterogeneous images, which have some additional knowledge that invariant in different light or pose conditions. In general, heterogeneous image are defined as different images from different sensors. In our database, we use both visual spectral (VIS) and near infrared (NIR) spectral for image capture, that’s why we call it a heterogeneous database. There are many other databases in the field of face including 2D VIS image database: YALE, FERET, FRGC; 3D image …