为了解决二维人脸识别准确度提升空间有限,三维人脸识别数据量大、识别速度慢的问题,提出了一种新的基于曲量场空间的人脸识别算法(Face Recognition based on Curved Space Field,FRCSF).该算法首先检测彩色人脸图像内的面部凸凹信息,...为了解决二维人脸识别准确度提升空间有限,三维人脸识别数据量大、识别速度慢的问题,提出了一种新的基于曲量场空间的人脸识别算法(Face Recognition based on Curved Space Field,FRCSF).该算法首先检测彩色人脸图像内的面部凸凹信息,利用曲量子描绘凸凹域的渐变梯度特征,去除人脸彩色信息,降低三维信息量.然后以分散的曲量子群融合成曲量子空间.将曲量子空间进行边缘曲量子光滑衔接,组成曲量场空间.最后提取曲量场空间内的深度和维度信息,通过与曲量人脸库进行信息对比,判别出人脸身份.该算法抓住了人脸面部的凸凹特征,继而将凸凹特征采用具有空间连续性规律约束的曲量场进行描述,识别准确率较高,同时由于对三维人脸采用曲量子进行重建,数据量小,识别速度较快.大量实验表明,该算法既保存了二维人脸识别速度快的长处,又融入了三维人脸识别的局部三维信息,具有较高的识别性能.展开更多
To achieve full-surface strain measurement of variable curvature objects,a 360°3D digital image correlation(DIC)system is proposed.The measurement system consists of four double-camera systems,which capture the o...To achieve full-surface strain measurement of variable curvature objects,a 360°3D digital image correlation(DIC)system is proposed.The measurement system consists of four double-camera systems,which capture the object’s entire surface from multiple angles,enabling comprehensive full-surface measurement.To increase the stitching quality,a hierarchical coordinate matching method is proposed.Initially,a 3D rigid body calibration auxiliary block is employed to track motion trajectory,which enables preliminary matching of four 3D-DIC sub-systems.Subsequently,secondary precise matching is performed based on feature points on the test specimen’s surface.Through the hierarchical coordinate matching method,the local 3D coordinate systems of each double-camera system are unified into a global coordinate system,achieving 3D surface reconstruction of the variable curvature cylindrical shell,and error analysis is conducted on the results.Furthermore,axial compression buckling experiment is conducted to measure the displacement and strain fields on the cylindrical shell’s surface.The experimental results are compared with the finite element analysis,validating the accuracy and effectiveness of the proposed multi-camera 3D-DIC measuring system.展开更多
文摘为了解决二维人脸识别准确度提升空间有限,三维人脸识别数据量大、识别速度慢的问题,提出了一种新的基于曲量场空间的人脸识别算法(Face Recognition based on Curved Space Field,FRCSF).该算法首先检测彩色人脸图像内的面部凸凹信息,利用曲量子描绘凸凹域的渐变梯度特征,去除人脸彩色信息,降低三维信息量.然后以分散的曲量子群融合成曲量子空间.将曲量子空间进行边缘曲量子光滑衔接,组成曲量场空间.最后提取曲量场空间内的深度和维度信息,通过与曲量人脸库进行信息对比,判别出人脸身份.该算法抓住了人脸面部的凸凹特征,继而将凸凹特征采用具有空间连续性规律约束的曲量场进行描述,识别准确率较高,同时由于对三维人脸采用曲量子进行重建,数据量小,识别速度较快.大量实验表明,该算法既保存了二维人脸识别速度快的长处,又融入了三维人脸识别的局部三维信息,具有较高的识别性能.
基金funded by the National Natural Science Foundations of China(Nos.12272176,U2037603).
文摘To achieve full-surface strain measurement of variable curvature objects,a 360°3D digital image correlation(DIC)system is proposed.The measurement system consists of four double-camera systems,which capture the object’s entire surface from multiple angles,enabling comprehensive full-surface measurement.To increase the stitching quality,a hierarchical coordinate matching method is proposed.Initially,a 3D rigid body calibration auxiliary block is employed to track motion trajectory,which enables preliminary matching of four 3D-DIC sub-systems.Subsequently,secondary precise matching is performed based on feature points on the test specimen’s surface.Through the hierarchical coordinate matching method,the local 3D coordinate systems of each double-camera system are unified into a global coordinate system,achieving 3D surface reconstruction of the variable curvature cylindrical shell,and error analysis is conducted on the results.Furthermore,axial compression buckling experiment is conducted to measure the displacement and strain fields on the cylindrical shell’s surface.The experimental results are compared with the finite element analysis,validating the accuracy and effectiveness of the proposed multi-camera 3D-DIC measuring system.