1. NO.5 South Street of Zhongguancun , Beijing China 10010
2. 电子工程系, 北京理工大学信息与电子学院 , Beijing China 100081
3. German Aerospace Center (DLR), Münchener Str. 20, 82234 Weßling , Wessling Germany 82234
4. Beijing Institute of Technology, NO.5 South Street of Zhongguancun , Beijing Beijing China 100081
In this paper, we propose to jointly use FrFT and CNN in a cascaded fashion, called FrFT convolutional face, which is capable of deeply extracting the richer facial features from both spatial domain and fractional Fourier domain. With the varied combination of different orders in FrFT and different components (e.g., amplitude and phase information), the proposed FrFT convolutional face is expected to learn the facial variations more effectively.
This research was supported, in part, by the National Natural Science Foundation of China under Grant 61331021 and Grant 61421001, and, in part, by the National Natural Science Foundation of China (U1833203).
Copyright 2020 Science China Press Co., Ltd. 《中国科学》杂志社有限责任公司 版权所有