Face Samples Re-lighting for Detection Based on the Harmonic Images

Different environment illumination has a great impact on face detection. In this paper, we present a solution by the face relighting based on the harmonic images. The basic idea is that there exist nine harmonic images which can be derived from a 3D model of a face, and by which we can estimate the illumination coefficient of any face samples. To detect faces under the certain lighting conditions, we relight those original face samples to get more new face samples under the various possible lighting conditions by an illumination ratio image and then add them to the training set. By train a classifier based on Support Vector Machine (SVM), the experimental results turn out that the relighting subspace is effective during the detection under the diverse lighting conditions. We also use the relighting database to train an AdaBoost-based face detector and test it on the MIT+CMU frontal face test set. The experimental results show that the data collection can be efficiently speeded up by the proposed methods.