Video-based face recognition has been widely studied in recently years. The common method of video face recognition is to recognize of every face frames that collected. The method not only cause large computing complexity, but also when face gesture and illumination change greatly, the face images collected contain very little information that can be used, these samples may have a negative effect to the results of recognition and even cause the recognition failure. Therefore, this paper proposes a method for facial video key frame extraction based on color histogram, the purpose is to remove the redundant frames, reduce the computational complexity and improve recognition efficiency.
展开▼