Multifarious distances, cameras and illuminations face database

Face recognition at a distance is still a challenging problem due to the low resolution face images resulting from the remote distance. To motivate researches on the problem and make up for the shortage of existing databases, we introduce MDCI database in this paper. The database contains 677 videos and 9734 pictures from 155 subjects captured by five different cameras, at four kinds of distances (1m, 3m, 5m, 7m), under two illuminations. Based on the evaluation protocol proposed in the paper, we tested baseline LBP and PCA algorithms on the database. Interested researchers can use the test results as a control performance when testing their own algorithms.