高吸收强度下的波长调制光谱

基于洛伦兹线型近似，研究了高吸收强度下波长调制吸收光谱中谐波信号的行为。得到了二、四、六次谐波幅度与吸收强度的关系的近似解析表达式。解析表达式的结果与数值计算得到的相符合。理论分析的结果在一个用于瓦斯传感的波长调制光谱系统上得到了验证。发现各偶次谐波随吸收强度的变化不仅有最大值还有零点，这一结论对瓦斯传感器的设计具有实际意义。

Lineshape Analysis of the Beat Signal Between Optical Carrier and Delayed Sidebands

This paper presents the lineshape analysis of the beat signal between the optical carrier and the shifted and delayed side-bands produced by sinusoidal amplitude modulation. It is shown that the beat signal has a typical lineshape with a very narrow delta-peak superposed on a quasi-Lorentzian profile. Theoretical explanation for the appearance of this peak has been given based on optical spectral structure constructed by a large number of optical wave trains. It is predicted that the delta-peak is originated from the beat between the wave trains in the carrier and those in the delayed sidebands when their average coherence length is longer than the delay line. Experiments carried out using different delay lines clearly show that the delta-peak is always located at the modulation frequency and decreases with the increasing delay line. Our analysis explicitly indicates that the linewidth is related to the observation time. It is also suggested that the disappearance of the delta-peak can be used as the criterion of coherence elimination.

Raman scattering of nanocrystalline silicon embedded in SiO2

Raman scattering of nanocrystalline silicon embedded in SiO2 matrix is systematically investigated. It is found that the Raman spectra can be well fitted by 5 Lorentzian lines in the Raman shift range of 100-600 cm(-1). The two-phonon scattering is also observed in the range of 600-1100 cm(-1) The experimental results indicate that the silicon crystallites in the films consist of nanocrystalline phase and amorphous phase; both can contribute to the Raman scattering. Besides the red-shift of the first order optical phonon modes with the decreasing size of silicon nanocrystallites, we have also found an enhancement effect on the second order Raman scattering, and the size effect on their Raman shift.

Degenerate four-wave-mixing signals from dc- and ac-driven semiconductor superlattice

Within the one-dimensional tight-binding model;rnd chi-3 approximation, we have calculated four-wave-mixing (FWM) signals for a semiconductor superlattice in the presence of both static and high-frequency electric fields. When the exciton effect is negligible, the time-periodic field dynamically delocalizes the otherwise localized Wannier-Stark states, and accordingly quasienergy band structures are formed, and manifest in the FWM spectra as a series of equally separated continua. The width of each continuum is proportional to the joint width of the valence and conduction minibands and is independent of the Wannier-Stark index. The realistic homogeneous broadening blurs the continua into broad peaks, whose line shapes, far from the Lorentzian, vary with the delay time in the FWM spectra. The swinging range of the peaks is just the quasienergy bandwidth. The dynamical delocalization (DDL) also induces significant FWM signals well beyond the excitation energy window. When the Coulomb interaction is taken into account, the unequal spacing between the excitonic Wannier-Stark levels weakens the DDL effect, and the FWM spectrum is transformed into groups of discrete lines. Strikingly, the groups are evenly spaced by the ac field frequency, reflecting the characteristic of the quasienergy states. The homogeneous broadening again smears out the line structures, leading to the excitonic FWM spectra quite similar to those without the exciton effect. However, all these features predicted by the dynamical theory do not appear in a recent experiment [Phys. Rev. Lett. 79, 301 (1997)], in which, by using the static approximation the observed Wannier-Stark ladder with delay-time-dependent spacing in the FWM spectra is attributed to a temporally periodic dipole field, produced by the Bloch oscillation of electrons in real space. The contradiction between the dynamical theory and the experiments is discussed. In addition, our calculation indicates that the dynamical localization coherently enhances the time-integrated FWM signals. The feasibility of using such a technique to study the dynamical localization phenomena is shown. [S0163-1829(99)10607-6].

Design and fabrication of polarization-insensitive 1550 mm semiconductor optical amplifiers

Polarization-insensitive semiconductor optical amplifiers (SOA's) with tensile-strained multi-quantum-wells as actice regions are designed and fabricated. The 6x6 Luttinger-Kohn model and Bir-Pikus Hamiltonian are employed to calculate the valence subband structures of strained quantum wells, and then a Lorentzian line-shape function is combined to calculate the material gain spectra for TE and TM modes. The device structure for polarization insensitive SOA is designed based on the materialde gain spectra of TE and TM modes and the gain factors for multilayer slab waveguide. Based on the designed structure parameters, we grow the SOA wafer by MOCVD and get nearly magnitude of output power for TE and TM modes from the broad-area semiconductor lasers fabricated from the wafer.

自主移动机器人三角定位的路标优化

针对给出的机器人三角定位算法，推导出路标设置对定位影响的几何精度标准，证明了参与定位的路标数目为3个和3个以上的情况下，如果路标与机器人的距离都相等，那么当相邻路标与机器人所成夹角都相等时，机器人的定位精度最高；进而给出了最优路标选取算法.实验和实践结果也进一步证明所得结论的有效性。

路标定位中的优化选择策略

基于合作路标的三角定位是自主式移动机器人的主要定位方法。路标的优化选择是提高定位算法精度的关键。本文在给出该算法原理的基础上 ,通过理论分析和实验证明了路标优化选择对提高定位算法精度的作用 ,并介绍了路标优化选择的准则

面向月面探测作业的双目立体视觉测量及三维环境生成技术研究

星球漫游车技术是一项跨学科的综合性技术，它反映了一个国家整体科技水平和高技术产业的发展水平，对漫游车进行研究具有长远的战略意义。漫游车一般都工作在特殊的环境中，综合考虑功耗和有效载荷等问题，可以发现，非视觉传感器与视觉传感器相比在环境感知能力和综合性能方面都不甚理想，因此，基于视觉传感器的立体视觉定位也一直是各类漫游车用以感知环境所首选的技术。基于立体视觉可以构建相应的地面人机交互系统。该交互系统具有实现离线示教功能，通过虚拟现实和立体视觉技术实现虚拟漫游车在虚拟地形中行走，虚拟机械臂对虚拟物体进行研磨等功能。操作人员在虚拟环境中对月球车的各种运动进行验证以后，再将数据上传给真实的漫游车以实现真实的漫游和考察工作。因此，立体视觉系统性能的好坏将直接影响到漫游车的后续工作情况，它是漫游车自主导航和进行科学考察的基础，直接关系到漫游车的安全，是一项关键的技术。 本文的研究依托于中国科学院知识创新工程方向性项目“月球车月面探测作业支撑技术研究”，旨在探索立体视觉系统的实现及其与车载机械臂配合工作的问题。从理论和实践两个方面，对其中的若干关键技术，如双摄像机标定、对极几何恢复、图像校正、图像匹配、三维重建和车载机械臂视觉定位等进行了研究。主要研究成果如下： 1．提出一种基于坐标排序的标定特征点存储算法，使得两幅图像中的标定特征点能够快速准确的对应存储。基于径向一致约束（RAC）模型优化方法与直接光学法的优点，提出一种改进的光心确定方法，实验表明该方法有效地提高了Tsai两步法的标定与测量精度。以上两种方法为实现双摄像机的高精度标定奠定了基础。 2．在双摄像机的标定阶段，提出一种新的以三维重投影误差为优化目标的双摄像机两步标定算法，实验证明该方法的标定与测量精度要高于传统的基于二维重投影误差的方法。提出一种考虑多个畸变系数基于三维重投影误差的两步标定算法，实验证明该方法比只考虑一个畸变系数的Tsai两步法能够获得更高的标定与测量精度。研究了基于遗传算法和粒子群算法的双摄像机内、外参数的整体优化问题，实验证明可以获得比传统梯度下降法更高的标定与测量精度。以上改进方法可以使标定精度达到工程要求，并为后续的图像校正和三维测量奠定基础。 3．从简单背景和复杂背景两种情况对对极几何的离线恢复问题进行了研究。对于对极几何的离线恢复问题，提出了基于正确匹配标定特征点进行基础矩阵高精度估计的算法。该算法的特点是与标定共用同一组数据，其有效性得到了实验的验证。对于对极几何的在线恢复问题，基于摄像机成像几何原理证明了图像坐标系下存在的一种左右图像坐标间的约束关系，并设计了相应的稀疏匹配算法，采用两阶段鲁棒稀疏匹配算法对基础矩阵进行了高精度的估计，实现了对极几何的在线恢复。与传统方法相比，本文的方法速度更快，精度更高。 4．研究了标定与非标定图像的极线校正算法，利用上述改进方法求得的高精度的投影矩阵和基础矩阵，获得了令人满意的校正效果，为稠密匹配奠定了坚实的基础。研究并实现了四种稠密匹配算法，并进行了算法性能的比较。针对稠密匹配后的初始候选匹配点对，研究了三种误匹配的剔除方法。实验结果表明这些方法能够有效剔除误匹配点对，提高三维重建的效果。 5．研究了基于最小二乘法和立体成像几何法的三维重建算法，研究了一种集成上述多种改进算法的欧式三维重建算法。研究开发了三维重建软件，基于真实图像进行了欧式三维重建实验研究，取得了良好的重建效果。 6．以月球车为应用对象，研究了虚拟机械臂和车载机械臂的平面定位机理，构建了基于立体视觉的虚拟机械臂定位仿真平台。基于虚拟环境下验证好的数据，进行了月球车车载机械臂平面定位的实验，结果证明车载机械臂具有良好的视觉定位精度。

一种快速获取激光拼焊焊缝错配的新算法

本文讨论了基于视觉和结构光的焊缝形貌视觉检测系统的组成原理,针对拼板激光焊接工程中对焊缝检测的实时性要求高的特点,创新性地提出了一种快速获得焊缝错配缺陷的算法,主要步骤为:首先基于有关标准的检测指标阈值的设定,再次,针对在线提取到的图像进行预处理,主要是加窗和中值滤波;最后为Radon变换与错配检测。该算法能减少计算任务,从而快速获得错配检测指标。实验给出了不等厚板拼焊时的线性错配的数值和分布,同时验证了该算法的有效性。

叠前地震数据规则化、重建及噪音压制

I address of reconstruction of spatial irregular sampling seismic data to regular grids. Spatial irregular sampling data impairs results of prestack migration, multiple attenuations, spectra estimation. Prestack 5-D volumes are often divided into sub-sections for further processing. Shot gathers are easy to obtain from irregular sampling volumes. My strategy for reconstruction is as follows: I resort irregular sampling gathers into a form of easy to bin and perform bin regularization, then utilize F-K inversion to reconstruct seismic data. In consideration of poor ability of F-K regularization to fill in large gaps, I sort regular sampling gathers to CMP and proposed high-resolution parabolic Radon transform to interpolate data and extrapolate offsets. To strong interfering noise--multiples, I use hybrid-domain high-resolution parabolic Radon transform to attenuate it. F-K regularization demand ultimately for lower computing costs. I proposed several methods to further improve efficiency of F-K inversion: first I introduce 1D and 2D NFFT algorithm for a rapid calculation of DFT operators; then develop fast 1D and 2D CG method to solve least-square equations, and utilize preconditioner to accelerate convergence of CG iterations; what’s more, I use Delaunay triangulation for weight calculation and use bandlimit frequency and varying bandwidth technique for competitive computation. Numerical 2D and 3D examples are offered to verify reasonable results and more efficiency. F-K regularization has poor ability to fill in large gaps, so I rearrange data as CMP gathers and develop hybrid-domain high-resolution parabolic Radon transforms which be used ether to interpolate null traces and extrapolate near and far offsets or suppress a strong interfere noise: multiples. I use it to attenuate multiples to verify performances of our algorithm and proposed routines for industrial application. Numerical examples and field data examples show a nice performance of our method.

地震勘探三维可视化方法研究与实现

This thesis mainly studies the technologies of 3-D seismic visualization and Graphic User Interface of seismic processing software. By studying Computer Graphics and 3-D geological modeling, the author designs and implements the visualization module of seismic data processing software using OpenGL and Motif. Setting seismic visualization flow as the subject, NURBS surface approximation and Delaunay Triangulation as the two different methods, the thesis discusses the key algorithms and technologies of seismic visualization and attempts to apply Octree Space Partitioning and Mip Mapping to enhance system performance. According to the research mentioned above, in view of portability and scalability, the author adopts Object-oriented Analysis and Object-oriented Design, uses standard C++ as programming language, OpenGL as 3-D graphics library and Motif as GUI developing tool to implement the seismic visualization framework on SGI Irix platform. This thesis also studies the solution of fluid equations in porous media. 2-D alternating direction implicit procedure has been turned into 3-D successive over relaxation iteration, which possesses such virtues as faster computing speed, faster convergence rate, better adaptability to heterogeneous media and less memory demanding.