作者单位
摘要
天津工业大学 人工智能学院, 天津 300387
相位提取与深度估计是结构光三维测量中的重点环节,目前传统方法在结构光相位提取与深度估计方面存在效率不高、结果不够鲁棒等问题。为了提高深度学习结构光的重建效果,本文提出了一种基于轻型自限制注意力(Light Self-Limited-Attention,LSLA)的结构光相位及深度估计混合网络,即构建一种CNN-Transformer的混合模块,并将构建的混合模块放入U型架构中,实现CNN与Transformer的优势互补。将所提出的网络在结构光相位估计和结构光深度估计两个任务上进行实验,并和其他网络进行对比。实验结果表明:相比其他网络,本文所提出的网络在相位估计和深度估计的细节处理上更加精细,在结构光相位估计实验中,精度最高提升31%;在结构光深度估计实验中,精度最高提升26%。该方法提高了深度神经网络在结构光相位估计及深度估计的准确性。
结构光 深度学习 自限制注意力 相位估计 深度估计 structured light deep learning self-limited attention phase estimation depth estimation 
中国光学
2024, 17(1): 118
作者单位
摘要
天津工业大学 天津市电气装备智能控制重点实验室,天津 300387
为了实现有特征物体和无特征物体更精准的三维重建,本文研究了多视角传感器下三维点云的自动拼接算法。首先由不同视角的传感器双目标定后进行轴线数据的标定,接着在三维空间内对多条轴线数据进行分析并提出了一种基于多视角传感器轴线融合的点云拼接方法,从而计算出误差最小的最优轴线数据,最后以拟合出的轴线数据为轴心在世界坐标系内进行三维点云的拼接。实验结果表明,在1.3~1.9 m的测量范围内,本文所提出的拼接方法对直径为144.954 2 mm的标准球进行三维重建的误差在0.037 mm以内,重建无特征点物体和有特征点物体都能有较好的拼接效果且拼接时间不受点云总量大小的限制。该拼接方法基本满足三维重建的稳定性好、效率快、精度高等要求。
三维重建 点云拼接 轴线 three-dimensional reconstruction point cloud splicing axis 
液晶与显示
2023, 38(6): 759
Author Affiliations
Abstract
1 Key Laboratory of Advanced Electrical Engineering and Energy Technology, Tiangong University, Tianjin 300387, China
2 School of Mechanical Engineering, Tianjin University of Technology and Education, Tianjin 300222, China
3 1Department of Precision Instrument, Tsinghua University, Beijing 100084, China
In order to realize the rapid detection of three-dimensional defects of connectors, this paper proposes a method for detecting connector defects based on structured light. This method combines structured light with binocular stereo vision to obtain three-dimensional data for the connector. Point cloud registration is used to identify defects and decision trees are used to classify defects. The accuracy of the 3D reconstruction results in this paper is 0.01 mm, the registration accuracy of the point cloud reaches the sub-millimeter level, and the final defect classification accuracy is 94%. The experimental results prove the effectiveness of the proposed three-dimensional connector defect detection method in connector defect detection and classification.
光电子快报(英文版)
2021, 17(2): 107
作者单位
摘要
1 天津工业大学天津市电工电能新技术重点实验室, 天津 300387
2 河北大学质量技术监督学院, 河北 保定 071002
可调谐激光吸收光谱技术(TDLAS)由于其高灵敏度、 高选择性等优势广泛用于痕量气体检测领域。 然而其测量结果容易受到目标气体压力波动的影响, 特别是在大气环境下尤为明显, 现有方法多为在现场安装压力传感器, 对测量结果进行校正。 提出了一种无需压力传感装置的气体浓度修正方法。 选取碰撞展宽占主导地位的气体吸收谱线, 分别建立谱线展宽与波长调制光谱一次谐波(WMS-1f)信号的峰谷值间距和二次谐波(WMS-2f)过零点间距的解析表达式, 通过测量一次谐波峰谷值间距或二次谐波过零点间距直接得到被测气体压强, 进而利用波长调制光谱一次谐波归一化的二次谐波(WMS-2f/1f)技术补偿测量环境中压力波动对气体浓度测量结果的影响。 实验以浓度为1 980 mg·m-3的CO2为目标气体, 选取其位于4 989.97 cm-1的吸收作为目标谱线, 在大气压附近进行不同调制深度的变压力测量实验, 通过实验分析了压强变化对二氧化碳吸收谱线谐波信号的影响, 利用一次谐波峰谷值间距和二次谐波过零点间距分别反演了气体压强, 并与气体压强传感器测得的压强数据进行对比, 压强偏差在1%以内, 验证了通过谐波间距解析表达式计算压强的正确性及通过测量谐波间距对浓度补偿的可行性。 最后利用WMS-2f/1f技术和通过谐波间距测得的压强数据对气体浓度进行压强补偿修正, 结果表明通过测量谐波间距修正后的浓度与通过高精度压力表补偿后浓度相比误差小于2%, 与通过谐波间距推导得出的压力不确定度(小于2%)一致, 验证了该方法的可行性和有效性, 进一步提高了TDLAS技术在压强波动较大环境下进行气体浓度检测的测量精度。 利用谐波间距对气体浓度补偿的方法无需额外的气体压力传感器, 简单易行, 特别适合于大气环境中气体成分的高灵敏高精度开放光路遥测, 也可用于气体浓度和压强的同时测量。
波长调制光谱 一次谐波 二次谐波 压强波动补偿 CO2检测 Wavelength modulation spectroscopy First harmonic signal Second harmonic signal Pressure fluctuation compensation CO2 detection 
光谱学与光谱分析
2020, 40(5): 1407
Author Affiliations
Abstract
1 Key Laboratory of Advanced Electrical Engineering and Energy Technology, Tianjin Polytechnic University, Tianjin 300387, China
2 School of Mechanical Engineering, Tianjin University of Technology and Education, Tianjin 300222, China
3 School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong NSW2500, Australia
This work deals with quantitative analysis of multicomponent mud logging gas based on infrared spectra. An accurate analysis method is proposed by combining a genetic algorithm (GA) and a radial basis function neural network (RBFNN). The GA is used to screen the infrared spectrum of the mixed gas, while the selected spectral region is used as the input of the RBFNN to establish a calibration model to quantitatively analyze the components of logging gas. The analysis results demonstrate that the proposed GA-RBFNN performs better than FS-RBFNN and ES-RBFNN, and our proposed method is feasible.
光电子快报(英文版)
2019, 15(4): 312
Author Affiliations
Abstract
Key Laboratory of Advanced Electrical Engineering and Energy Technology, Tianjin Polytechnic University, Tianjin 300387, China
The prediction method of dynamic wavelength is proposed for temperature tuning process. The temperature of the thermistor integrated in laser diode (LD) module is recorded to predict the LD chip temperature. Then according to the injection current and priori tuning characteristics of the LDs, the emission wavelength is estimated in real time. The method is validated by using a 1.58 μm distributed feedback (DFB) LD. The absorption spectra of mixture gas of CO2and CO are measured by means of the thermal tuning gas sensing system. The center wavelength of each absorption line is compared with the data in HITRAN2012 database. The results show that the deviations are less than 5 pm. This method fully meets the needs of spectroscopic measurement, and can be applied to spectroscopy, optical communications and other fields.
光电子快报(英文版)
2017, 13(2): 108
作者单位
摘要
1 天津工业大学天津市电工电能新技术重点实验室, 天津 300387
2 天津大学精密测试技术及仪器国家重点实验室, 天津 300072
空芯光波导(HWG)用于光谱气体检测中, 既可以实现光路的传输, 又可以充当气体样品池实现长光程高灵敏度测量, 具有体积小, 响应时间快、 成本低、 光路稳定灵活等优点。 介绍了基于镀银/碘化银的空芯光波导(Ag/AgI-HWG)、 光子带隙空芯光波导(PBG-HWG)和基片集成空芯光波导(iHWG)等类型的空芯光波导, 并总结了近年来空芯光波导在光谱气敏检测中的研究及进展, 梳理了其应用方式及应用领域。 研究表明, 空芯光波导替代传统的气体池与傅里叶变换红外光谱(FTIR)、 激光吸收光谱和拉曼光谱等不同的光谱技术结合已取得一系列成果, 且已经应用于环境监测、 呼气诊断和工业过程检测和控制等领域。 其中, 基于中红外激光吸收光谱的空芯光波导传感器组成相对简单, 成本较低, 与各类光波导的兼容性和环境适应性较强, 发展前景较好。 总之, 随着激光技术、 光波导技术和光谱技术的发展, 基于空芯光波导的光谱气体检测正在迅速发展, 并逐步由实验室走向现场应用。
空芯光波导 傅里叶变换红外光谱 激光吸收光谱 拉曼光谱 气体检测 Hollow waveguide FTIR Laser absorption spectroscopy Raman spectroscopy Gas detection 
光谱学与光谱分析
2017, 37(7): 2259
Author Affiliations
Abstract
1 Key Laboratory of Advanced Electrical Engineering and Energy Technology, Tianjin Polytechnic University, Tianjin 300387, China
2 Tianjin University of Technology and Education, Tianjin 300222, China
3 School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong 2500, Australia
An optical hydrogen sulfide (H2S) sensor based on wavelength modulation spectroscopy with the second harmonic (2f) corrected by the first harmonic (1f) signal (WMS-2f/1f) is developed using a distributed feedback (DFB) laser emitting at 1.578 μm and a homemade gas cell with 1-m-long optical path length. The novel sensor is constructed by an electrical cabinet and an optical reflecting and receiving end. The DFB laser is employed for targeting a strong H2S line at 6 336.62 cm-1in the fundamental absorption band of H2S. The sensor performance, including the minimum detection limit and the stability, can be improved by reducing the laser intensity drift and common mode noise by means of the WMS-2f/1f technique. The experimental results indicate that the linearity and response time of the sensor are 0.999 26 and 6 s (in concentration range of 15.2—45.6 mg/m3), respectively. The maximum relative deviation for continuous detection (60 min) of 30.4 mg/m3H2S is 0.48% and the minimum detection limit obtained by Allan variance is 79 μg/m3with optimal integration time of 32 s. The optical H2S sensor can be applied to environmental monitoring and industrial production, and it has significance for real-time online detection in many fields.
光电子快报(英文版)
2016, 12(6): 465
Author Affiliations
Abstract
1 Key Laboratory of Advanced Electrical Engineering and Energy Technology, Tianjin Polytechnic University, Tianjin 300387, China
2 School of Electrical, Computer and Telecommunications Engineering, University of Wollongong, Wollongong 2500, Australia
In order to get the entire data in the optical measurement, a multi-view three-dimensional (3D) measurement method based on turntable is proposed. In the method, a turntable is used to rotate the object and obtain multi-view point cloud data, and then multi-view point cloud data are registered and integrated into a 3D model. The measurement results are compared with that of the sticking marked point method. Experimental results show that the measurement process of the proposed method is simpler, and the scanning speed and accuracy are improved.
光电子快报(英文版)
2016, 12(5): 389
作者单位
摘要
1 天津工业大学 电气工程与自动化学院, 天津 300387
2 天津大学 电子信息工程学院, 天津 300072
针对条纹投影三维形貌测量涉及的相位提取, 提出了一种基于变分模态分解的单幅条纹投影相位提取方法。通过建立变分模态分解模型和极小化变分模态分解将单幅投影条纹图分解成背景部分、条纹部分和噪声部分。然后对得到条纹部分进行Hilbert变换和反正切变换得到包裹相位; 对其进行质量导向相位解包裹和Zernike多项式去载频得到解包裹相位。将该方法与Fourier变换、连续小波变换进行了对比, 结果显示: 本文提出的相位提取方法相位误差为3.14×10-4, 小于Fourier变换和连续小波变换方法对应的误差3.30×10-4和6.52×10-4。模拟和实验结果表明: 本文提出的方法在处理具有边缘信息投影条纹图时具有优势, 能够提取出更准确的相位信息, 可有效地用于含边缘不连续和突起的三维物体测量。
条纹投影 三维形貌测量 变分模态分解 相位提取 fringe projection three dimensional measurement variational mode decomposition phase retrieval 
光学 精密工程
2016, 24(9): 2318

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!