부산대학교 도서관

In this paper, we propose a novel face recognition method based on fusing the near infrared and visible images of face images with distributed compressive sensing. The near infrared image and visible image of one same subject constitute an ensemble. Both images in one ensemble share a common sparse component while each individual image has an innovation component. To better capture the complementary information of the ensemble, the distributed compressive sensing is used to obtain the common component and the innovation component of near infrared and visible image. The obtained common component contains the complementary information of near infrared and visible image effectively. So the sparse coefficients of the common component obtained by distributed compressive sensing can better capture the intrinsic structures of each image and therefore can obtain better performance than that of only using near infrared image or visible image. The experimental results on several benchmark datasets demonstrate the effectiveness of proposed method.