Raman lasers are essential in atomic physics,and the development of portable devices has posed requirements for time-division multiplexing of Raman lasers.We demonstrate an innovative gigahertz frequency hopping appro...Raman lasers are essential in atomic physics,and the development of portable devices has posed requirements for time-division multiplexing of Raman lasers.We demonstrate an innovative gigahertz frequency hopping approach of a slave Raman laser within an optical phase-locked loop(OPLL),which finds practical application in an atomic gravimeter,where the OPLL frequently switches between near-resonance lasers and significantly detuned Raman lasers.The method merges the advantages of rapid and extensive frequency hopping with the OPLL’s inherent low phase noise,and exhibits a versatile range of applications in compact laser systems,promising advancements in portable instruments.展开更多
针对人脸识别准确率易受人脸表情变化影响的问题,提出了一种分区域特征提取的人脸识别算法。首先,在预处理图像上标记出14个人脸关键点,并将人脸图像划分为表情易变区域和不变区域;然后,分别用Gabor+LBP和Gabor+分块LBP两种特征提取通...针对人脸识别准确率易受人脸表情变化影响的问题,提出了一种分区域特征提取的人脸识别算法。首先,在预处理图像上标记出14个人脸关键点,并将人脸图像划分为表情易变区域和不变区域;然后,分别用Gabor+LBP和Gabor+分块LBP两种特征提取通道对表情不变区域和易变区域进行特征提取;最后,将所得的特征直方图级联,并进行身份验证。经FERET(face recognition technology),LFW(labled faces in the wild)及自制人脸数据库验证,文中算法准确率分别达到了99.14%,98.5%,96.52%。在FERET数据库中,该文算法准确率较DeepID和Gabor+分块LBP算法分别提高了1.88%和3.6%,F1(调和平均数)分别提高了1.8%和2.86%。实验结果表明,分区域特征提取的人脸识别算法对人脸表情变化具有很强的鲁棒性。展开更多
基金Project supported by the National Key Research and Development Program of China(Grant Nos.2021YFA0718300 and 2021YFA1400900)the National Natural Science Foundation of China(Grant Nos.11920101004,11934002,and 92365208)+1 种基金Science and Technology Major Project of Shanxi(Grant No.202101030201022)Space Application System of China Manned Space Program.
文摘Raman lasers are essential in atomic physics,and the development of portable devices has posed requirements for time-division multiplexing of Raman lasers.We demonstrate an innovative gigahertz frequency hopping approach of a slave Raman laser within an optical phase-locked loop(OPLL),which finds practical application in an atomic gravimeter,where the OPLL frequently switches between near-resonance lasers and significantly detuned Raman lasers.The method merges the advantages of rapid and extensive frequency hopping with the OPLL’s inherent low phase noise,and exhibits a versatile range of applications in compact laser systems,promising advancements in portable instruments.
文摘针对人脸识别准确率易受人脸表情变化影响的问题,提出了一种分区域特征提取的人脸识别算法。首先,在预处理图像上标记出14个人脸关键点,并将人脸图像划分为表情易变区域和不变区域;然后,分别用Gabor+LBP和Gabor+分块LBP两种特征提取通道对表情不变区域和易变区域进行特征提取;最后,将所得的特征直方图级联,并进行身份验证。经FERET(face recognition technology),LFW(labled faces in the wild)及自制人脸数据库验证,文中算法准确率分别达到了99.14%,98.5%,96.52%。在FERET数据库中,该文算法准确率较DeepID和Gabor+分块LBP算法分别提高了1.88%和3.6%,F1(调和平均数)分别提高了1.8%和2.86%。实验结果表明,分区域特征提取的人脸识别算法对人脸表情变化具有很强的鲁棒性。