基于视场拼接的全视差三维显示系统

提出了一种新的基于视场拼接的全视差三维显示方法及图像生成方案.系统核心部分由一个微投影机阵列、一个反射光路和一个散射屏组构成,图像生成方案采用基于OpenGL的算法实现.为了验证所提出的基于视场拼接理论的正确性,利用液晶显示器、菲涅耳镜组、定向散射屏和背投式反射光路搭建了实验样机.该装置可以在更紧凑的系统结构下显示包括横向视差和纵向视差在内的三维图像,并且通过图像生成算法的配合实现了动态三维显示效果.     

基于双层微透镜阵列的移动视差式立体显示屏

受限于微透镜的数值孔径和像差问题,目前移动视差式立体显示屏存在多视场空间、视场区之间存在视场跳变、主视场空间角度小、显示效果模糊等问题.本文基于离散采样思想,分析了移动视差型三维显示系统的工作机理,结合加工工艺,设计了复合曲面反射层的双层微透镜阵列屏.仿真结果表明:所设计显示屏的视场观看空间角可提升到60°,相对于传统的三维显示屏,双层微结构显示屏可有效提高视场角和显示性能.     

周视悬浮立体显示系统的光学系统设计

基于双目视差原理,提出了一种新型的周视悬浮裸眼立体显示系统。采用菲涅尔透镜和单向散射屏实现了观察范围的纵向拓展,形成的三维图像可周视观看,有悬浮感且可触摸。对系统中采用的定向屏、旋转光阑、条状出瞳,以及投影机的投影距离进行了分析和设计。经过实验验证,立体图像合像舒适,显示效果明显。     

基于变形目镜和OLED的全景周视成像与显示技术

在车载、舰载等运动载体或单兵警戒值守等狭窄空间,传统显示方法难以兼顾显示视场与显示分辨率。针对该问题,本文研究了一种基于变形目镜的全景/周视成像技术与系统。系统采用3路低照度摄像机+4 mm焦距镜头,构成150°左右的全景成像视场,基于FPGA处理平台完成全景图像解析、拼接、校正和显示等算法,实现视频图像大视场拼接与1/3缩放实时显示;由OLED微显示器、变形目镜组和大目镜组成的显示系统使视频图像横向扩大3倍,实时显示全景高分辨动态场景图像。实验系统对全景成像显示技术进行了验证,在军用和民用领域具有广泛的应用前景。     

一种适用于仿生复眼系统的多路图像拼接算法研究

仿生复眼系统解决了传统单孔径光学系统中视场与分辨率的矛盾关系,兼具了较大的视场和高分辨率。但随着视场和分辨率的增加,图像信息增多,随之带来了图像拼接中效率和质量问题。针对该问题,提出一种基于改进尺度不变特征转换（SIFT）算法和主成分分析（PCA）算法的仿生复眼多路图像拼接融合方法。该方法缩小了特征点提取区域,减少了多路图像特征点匹配次数,降低了图像特征点描述符的维度。再利用改进的自适应迭代随机抽样一致（RANSAC）算法对特征点进行提纯增加鲁棒性,最后通过加权平均算法来完成对多幅子图像的高质量融合。实验结果表明,该算法设计合理,且随着图像信息的复杂程度增加,相较传统算法的拼接效率不断提升,同时保证了较好的拼接质量,可有效拼接融合仿生复眼系统多路图像,对多路图像类系统拼接融合提供借鉴。     

基于SIFT匹配算法的多视点自由立体显示视差图像的生成

提出了一种利用同一场景的两幅视差图像生成用于多视点自由立体显示的多幅视差图像的方法。首先运用SIFT匹配算法从两幅视差图像中寻找匹配点,并运用生长运算得到图像的稠密匹配;然后根据匹配点坐标对视差进行了近似计算并得到左视差图像的深度图像;最后提出一种基于投影原理的多幅视差图像生成方法,由左视差图像和其深度图像生成了应用于多视点自由立体显示的多幅视差图像,其立体显示效果良好。     

大视场复眼结构图像处理算法研究

针对多维子眼成像通道曲面排布组成的大视场复眼结构,提出了一种切割-旋转-映射的图像处理算法来实现多通道图像的大视场拼接.通过确定复眼结构的排布特征,分析了各成像通道捕获的子眼图像之间的相互关系,去除相邻子眼图像之间的冗余部分,并运用几何光学及成像光学原理,研究了子眼图像与三维映射空间之间的关系,从而实现了二维子眼图像在三维空间的大视场拼接.实验制备了包含37个镜头且视场角可达118°的人工复眼结构,并运用提出的图像处理算法处理制备的复眼结构捕获的子眼图像.结果表明:算法处理图像过程中不损失图像的分辨率,可以有效地实现多通道图像的大视场拼接,且获得的图像可视性强,满足实用化要求,可进一步推进曲面复眼成像系统的应用.     

基于片间相关性的光线空间插值与视点绘制研究

提出了基于片间相关性的光线空间插值算法,利用邻近参考片已知视差,快速预测当前片视差,使得当前片视差估计的搜索限制在预测视差附近的小范围内进行,从而大大降低了计算复杂度.根据视差参考方式的不同,设计了三种实现方案.实验结果表明,采用双向视差参考方案插值生成的光线空间片不但信噪比高、绘制视点图像主客观质量较高,而且插值与绘制速度也有了明显提高.     

基于大视场人工复眼定位技术

针对大视场目标探测提出了一种基于人工复眼大视场定位方法.通过分析子眼视场角与总视场角之间的关系,并结合多目视觉定位对子眼排布方式的要求,研究了包含多个子眼的人工复眼结构设计方法.通过分析子眼图像与三维空间映射关系,对二维图像进行裁剪并映射于三维立体空间,实现了二维子眼图像在三维空间的大视场拼接.利用子眼图像坐标、空间三维坐标及系统参数间的关系,建立了空间点多目定位数学模型,并编制目标定位算法.制备了包含19个子眼可实现120°大视场角的样机,通过张正友标定法获得系统参数,并进行目标定位实验.实验结果表明,使用设计的人工复眼大视场成像系统对5.35m处目标进行探测,定位误差为0.19%.     

基于柔性靶标定位实现图像拼接的多相机三维测量系统

提出一种基于柔性靶标定位实现图像拼接的多相机三维测量系统,采用一个激光投影仪投影大幅条纹,多相机分布式采集的方法扩展视觉三维测量系统的测量范围。标定过程首先使用小型平面靶标标定基准相机二维图像坐标和相位值到三维世界坐标的映射关系,之后在相邻相机部分视场(FOV)重合的前提下,利用柔性靶标定位标定相邻相机图像坐标的转换关系,最后将各个相机的图像坐标全部转化到基准相机的图像坐标系下完成图像拼接,由基准相机图像坐标到世界坐标的映射完成全局三维测量。实验结果表明,使用图像拼接方法的测量精度略低于相机单视场测量的精度,但精度损失较小,满足工业在线测量的要求。该方法避免使用昂贵的辅助测量仪器和加工高精度大型靶标,为多相机视觉测量提供了成本低、使用方便的解决方案。     

Aberration improvement of the floating 3D display system based on Tessar array and directional diffuser screen

The floating 3D display system based on Tessar array and directional diffuser screen is proposed. The directional diffuser screen can smoothen the gap of lens array and make the 3D image’s brightness continuous. The optical structure and aberration characteristics of the floating three-dimensional (3D) display system are analyzed. The simulation and experiment are carried out, which show that the 3D image quality becomes more and more deteriorative with the further distance of the image plane and the increasing viewing angle. To suppress the aberrations, the Tessar array is proposed according to the aberration characteristics of the floating 3D display system. A 3840?×?2160 liquid crystal display panel (LCD) with the size of 23.6 inches, a directional diffuser screen and a Tessar array are used to display the final 3D images. The aberrations are reduced and the definition is improved compared with that of the display with a single-lens array. The display depth of more than 20 cm and the viewing angle of more than 45° can be achieved.     

Full-parallax three-dimensional display using new directional diffuser

A full-parallax three-dimensional(3D) display using a new directional diffuser is demonstrated.The display could present 3D images with 45 views comprising 9 horizontal views by 5 vertical views.The resolution and size of the 3D images displayed are 226×226 pixels and 300×300(mm).The new directional diffuser consisting of two perpendicular lenticular sheets can be widely used in the display domain owing to its low cost and simple process.     

Shared pixel based parallax barrier 3D display with high brightness

The parallax barrier 3D display is an autostereoscopic display with low cost. However, the conventional parallax barrier 3D display has a disadvantage of low brightness. Therefore, a parallax barrier 3D display based on shared pixels is proposed. The shared pixels have adapted pixel values which limit crosstalk in a low level, so that they can be displayed simultaneously in both two parallax images. These pixels increase aperture ratio of the parallax barrier. Therefore, they can bring high brightness. A prototype of the proposed 3D display is developed and it has higher brightness than those based on the conventional one.     

Combined lenticular lens for autostereoscopic three dimensional display

When an autostereoscopic three dimensional (3D) display based on a single lenticular lens keeps working for a long period, the thermal expansion of lenticular lens results in serious crosstalk and 3D image deterioration. To overcome this problem, a combined lenticular lens is proposed. The combined lenticular lens is composed of a single lenticular lens used in the traditional autostereoscopic 3D display, a transparent glue layer and a glass substrate. The design method of the combined lenticular lens and analysis of thermal expansion effect is given. A 42-inch autostereoscopic 3D display based on the combined lenticular lens is developed. The experiment results verify that the proposed combined lenticular lens effectively avoids the crosstalk and 3D image deterioration happened in the traditional autostereoscopic 3D display.     

A ferroelectric liquid crystal spatial light modulator encoded with orthogonal arrays and its optimized design for laser speckle reduction

In laser projectors, speckle reduction can be achieved by projecting a changing binary phase diffuser onto the screen. Here, we sequentially encoded a commercialized ferroelectric liquid crystal spatial light modulator (FLC-SLM) with the rows of a two-level orthogonal array of order sixty-four, thus obtaining a changing binary phase diffuser. With the help of this binary phase diffuser, the subjective speckle contrast ratio on the screen is reduced from C_b=0.71 to C_a=0.1. Based on the experimental results, a simplified transparent FLC-SLM design is first proposed. This newly designed FLC-SLM has two phase modulation depths and can be driven with the passive matrix addressing scheme. Therefore, the control electronics of the proposed FLC-SLM can be significantly simplified compared to the currently used one.     

一种符号阵列编码结构光三维检测方法

编码结构光检测技术是一种主动视觉方法,利用投射的模式光,通过三角原理获得深度信息。根据伪随机阵列的特性,提出了一种可用于检测场景目标的一次投射模式的结构光三维检测方法。符号的拐点、交叉点比传统的基于伪随机序列及M阵列模式提供了更多的检测点。符号阵列能够提供足够多的码字使所有的子模式获得全局唯一性。基于符号链分解算法及角度变化的轮廓特征,解码方法能够识别绝大多数码字。重建实验表明可以对一定曲面变化的物体进行检测,并能重建其形廓。     

燃气透平末级叶片及扩压器的联合优化

本文基于iSIGHT优化软件包以及叶轮机械三维粘性CFD数值模拟软件包,采用包括全局探索式、梯度搜索 式优化算法以及近似模型的组合优化策略,对某燃气透平的末级叶型和扩压器外形进行了联合优化。优化后的透平末级与 扩压器部分的CFD预测效率提高了2.8％,同时扩压器的长度缩短了30.1％。     

基于光场重构的空间三维显示技术

真实空间三维显示可以通过精确再现三维物体的空间光场分布来实现.依据三维景物空间光场分布特性,提出并介绍基于光场重构的真实空间三维场景显示的基本原理与方法.实验表明,利用光场重构原理,可以运用现有的光学空间光调制器构造出比全息再现更为优越的真实三维显示.     

Fraunhofer computer-generated hologram for diffused 3D scene in Fresnel region

A Fraunhofer computer-generated hologram (CGH) is proved to be valid in display for three-dimensional (3D) objects from the Fresnel to the far-field region without a Fourier lens for reconstruction. To quickly compute large and complicated 3D objects that consist of slanted diffused surfaces in the Fresnel region, a Fraunhofer-based analytical approach using a basic-triangle tiling diffuser is developed. Both theoretical and experimental results reveal that Fraunhofer CGH can perform the same effects as Fresnel CGH but require less calculation time. Impressive 3D solid effects are achieved in the Fresnel region.