未知室内环境的三维景物描述

景物描述是计算机视觉的一个重要内容。提出了一个利用三维信息对未知景物的描述方法。该方法对三维场景进行物体级和结构级两级描述；根据三维物体分割的结果，首先计算单个物体的最大尺度、形状类型和边集等整体特征，继而推测它们之间的相互位置关系。从数学上推导了“简化边框矩”和“平面惯量张量”的计算公式，并给出了实验结果。     

海面目标高分辨率卫星成像仿真方法

海洋目标高空间分辨率遥感成像仿真技术在海面目标探测识别等方面得到了广泛应用。舰船与海水流体交互作用在高分辨率下得以显现,对其产生的复杂流场辐射模拟是成像仿真的主要难点。重点研究了舰船航行过程中与海水交互产生的流场几何形态和物性变化,提出了与海面方向辐射特性的耦合作用模型及海洋目标高分辨率遥感成像仿真方法。通过频谱分析的方法构建海面三维模型,使用计算流体力学的方法构造了船只航行流场的三维几何模型。根据海面组分分布的不同将其辐射特性与三维结构关联,构建了亚米级海洋场景三维辐射模型。通过辐射传输计算、场景内部多次反射模拟及大气影响和传感器效应仿真,最终得到观测条件下的卫星遥感图像。结果表明,将GF-6卫星全色波段实测图像与相同成像条件下的仿真图像对比,图像均值的误差为9.17%,标准差误差为9.21%,在平均灰度值、灰度分布、纹理细节等方面都具有较好的一致性,可以较真实地模拟高分辨率卫星成像下的海洋目标场景。     

基于二值神经网络的大场景点云分类

近年随着3维数据采集技术不断发展,大场景 点云数据的获取越来越方便。目前深 度学习网络框架在2维图像处理领 域越来越成熟,而大场景点云是一种3维无规则化的数据,3维卷积神经网络直接处理大场 景3维数据会存在分类精度低和计 算复杂等问题。因此为了有效解决基于深度学习的点云分类任务中存在的计算时间长和分类 精度低的问题,本文提出基于二值 神经网络的大场景点云分类方法,针对不规则的3维点云数据设计特征值计算方法,基于IR -Net二值神经网络处理输入的点云 特征图像,进一步采用Dynamic ReLU激活函数,提高神经网络的计算效率,最后得出点云分 类结果。实验结果表明,所提出 的方法在Oakland数据集上分类精度达到97.6%,在GML数据集中取得 了92.3%和97.2%的分类精度,实验结果证明Dy -ResNet 能够有效提升了点云分类的精度,减少计算的复杂度,并提高了训练效率。     

使用双传感器系统降低光刀图像缺损误差

用线结构光扫描测量三维物体的过程中,由于物体表面的凹凸不平、反射率的不连续变化以及遮挡造成获取的条纹像存在各种缺陷,导致最后得到的三维面形数据存在偏差。基于此,提出了一种采用双传感器,并将两个传感器图像映射到光刀平面上,然后再对其进行面内融合处理最后提取中心线的方法。对此方法进行了计算机模拟,并对牙齿进行了测量。实验结果表明该方法实现简单,在存在条纹缺陷的地方,精度可以明显提高,有效地降低了条纹缺陷导致的数据偏差。     

高分辨率卫星海洋背景成像仿真方法

海洋背景的仿真是海面目标场景仿真的关键环节,是海洋目标-背景耦合作用模拟的重要基础,决定了仿真图像中目标与背景差异特征的正确性和真实性。高分辨率遥感成像下,海面细节特征突显,以往视海表为均匀辐射面的处理方法给高分辨率海洋场景仿真造成较大误差。重点研究了海面三维形态、多组分分布与海水方向反射特性的耦合作用和辐射模型,提出了海面高分辨率遥感成像仿真方法。通过频谱分析的方法构建海面三维模型,根据海面组分分布的不同及不同位置海面法向的不同,修正了低分辨率下的海洋BRDF模型,使其满足高分辨率卫星图像的仿真应用,计算出不同组分的海面的方向反射数据,并将其与海面三维模型关联,构建了亚米级海面三维辐射模型,通过光线追踪方法建立海面零视距辐射场,并经过大气影响和传感器效应,模拟不同海况条件下卫星遥感图像。结果表明:将ZY3-02卫星实测海面图像与相同成像条件下的仿真图像对比,图像均值的误差为3.7%,标准差误差为9.9%,可以较真实地模拟高分辨率卫星成像下的海洋背景。     

Image segmentation through graph-based clustering from surface normals estimated by photometric stereo

A method for segmenting 2D images based on 3D shape information is proposed. First, a robust photometric stereo technique estimates the 3D normals of the objects present in the scene for every image pixel. Then, the image is segmented by grouping its pixels according to their estimated normals through graph-based clustering. Differently from other image segmentation algorithms based on intensity, colour or texture, the regions of which are determined by the visual appearance of the depicted objects, the regions obtained with the proposed technique only depend on the 3D shapes of those objects. This can be advantageous for higher level scene understanding algorithms. This technique is especially suited to poorly illuminated scenarios and utilises a conventional camera and six inexpensive strobe lights.     

基于学习的光栅图像噪声抑制方法

基于条纹投影的三维形貌测量广泛应用于工业制造、质量检测、生物医疗、航空航天等领域。然而在高速测量的场景下,由于光栅图像的采集过程曝光时间短,三维重建结果通常会受到较为严重的图像噪声干扰。近年来,深度学习技术在计算机视觉等领域得到了广泛应用,并且取得了巨大的成功。受此启发,提出了一种基于学习的光栅图像噪声抑制方法。首先构建了一个基于U-net的卷积神经网络。其次在训练过程中,构建的神经网络学习从含有噪声的条纹图像到对应高质量包裹相位之间的映射关系。当经过适当训练,该网络可从含有噪声的条纹图像中准确恢复相位信息。实验结果表明:针对离线的快速运动场景三维测量,该方法仅利用一幅光栅图像可恢复高精度的相位信息,且相位精度优于传统的三步相移方法。该方法可为提升运动高速场景三维测量的精度提供切实可靠的解决方案。     

基于空间截面投影的 Otsu 图像分割算法

针对传统三维Otsu法存在抗噪性差、计算复杂度高、难以多阈值扩展等不足,提出了一种基于空间截面投影的Otsu法。该方法采用垂直于主对角线的平面截三维直方图,并将截面内的点投影到主对角线上,进而建立一维的截面投影直方图以降低计算复杂度,然后基于该一维直方图进行Otsu分割;对于分割结果,运用基于阈值的后处理策略做进一步的处理以增强算法的抗噪性。实验结果表明：与现有三维Otsu法相比,该算法计算效率高、抗噪性好,能对含不同噪声的图像进行较好的分割。     

Autocorrelation and regularization in digital images. II. Simpleimage models

For pt.I see ibid., vol.26, no.4, p.463-73, July 1988. The variogram function used in geostatistical analysis is a useful statistic in the analysis of remotely sensed images. Using the results derived in Part I, the basic second-order, or covariance, properties of scenes modeled by simple disks of varying size and spacing after imaging into disk-shaped pixels are analyzed to explore the relationship between the image variograms and discrete object scene structure. The models provide insight into the nature of real images of the Earth's surface and the tools for a complete analysis of the more complex case of three-dimensional illuminated discrete-object images     

基于人工智能的足迹识别与特征提取

针对战场感知及侦破现场中传统人工主观经验检验与识别模式误差较大的问题,提出了一种基于人工智能的足迹识别与特征提取方法。采用三维形貌重构系统进行足迹图像采集,并将数字图像处理算法与传统足迹检验法结合,提取足迹的区域关系特征和形状长度特征,进而采用支持向量机的模式识别方法对提取的特征进行立体足迹身份鉴别对比实验。实验结果表明,所提方法准确率超过人工鉴别准确率,达到99.1%,可应用于战场感知及侦破现场足迹准确检测与识别,也可推广应用于人体身份鉴别的相关领域。     

Computational Reconstruction of Three-Dimensional Integral Imaging by Rearrangement of Elemental Image Pixels

In this paper, we present a method to implement computational three-dimensional (3D) integral imaging (II). This method is based on Pixels of the Elemental Image Rearrangement Technique (PERT). In our proposed method for computational reconstruction of II, the reconstructed 3D image is obtained by using the entire elemental images which are captured from the lenslet array. Instead of averaging the elemental images, our proposed method rearranges pixels of each elemental image. Therefore, the reconstructed 3D image has the same number of pixels as the entire elemental images' pixels. To verify this computational reconstruction method, we have implemented optical experiments.      

基于激光与可见光同步数据的室外场景三维重建

本文根据TH-3DLCS-2001三维成像扫描仪可同步扫描激光与可见光数据的特点,提出了室外场景三维重建的系统方法.推导了激光与可见光采样点的配准关系,给出了每一激光采样点的RGB值;引入马氏距离,提出了基于反向投影的多机位三维数据快速配准算法,将多机位三维数据无缝拼合在同一坐标系内;依据Hough变换和模糊子集,划分室外三维数据场为房顶、墙面、窗户、地面、树木、其他物体等6类,并给出了网格剖分方法.实验证明本文所提出的方法有效地解决了室外场景三维重建的基本问题.     

Model-based quantitation of 3-D magnetic resonance angiographic images

Quantification of the degree of stenosis or vessel dimensions are important for diagnosis of vascular diseases and planning vascular interventions. Although diagnosis from three-dimensional (3-D) magnetic resonance angiograms (MRA's) is mainly performed on two-dimensional (2-D) maximum intensity projections, automated quantification of vascular segments directly from the 3-D dataset is desirable to provide accurate and objective measurements of the 3-D anatomy. A model-based method for quantitative 3-D MRA is proposed. Linear vessel segments are modeled with a central vessel axis curve coupled to a vessel wall surface. A novel image feature to guide the deformation of the central vessel axis is introduced. Subsequently, concepts of deformable models are combined with knowledge of the physics of the acquisition technique to accurately segment the vessel wall and compute the vessel diameter and other geometrical properties. The method is illustrated and validated on a carotid bifurcation phantom, with ground truth and medical experts as comparisons. Also, results on 3-D time-of-flight (TOF) MRA images of the carotids are shown. The approach is a promising technique to assess several geometrical vascular parameters directly on the source 3-D images, providing an objective mechanism for stenosis grading.     

时序平滑多尺度叠加动态红外云场景仿真

天基红外观测场景中云场景具有几何结构动态变化复杂、尺度变化随机、辐射动态变化不确定的特点且与天基动态探测链路耦合,对系统探测效能造成极大影响。因此,开展云场景仿真方法研究对天基红外光学卫星系统设计具有非常重要的意义。针对传统仿真方法在大尺度动态云图像仿真应用中计算效率低、占用内存大的问题,提出了一种基于时序平滑多尺度叠加方法的动态云图像仿真方法,在基于多尺度叠加算法进行云层整体位置移动仿真的基础上,利用帧间插值的方法实现动态云层形状结构变化,计算效率提高10倍以上,真实模拟云场景的整体移动与内部变化,实现大尺度动态云图像的仿真。     

Fully 3-D PET reconstruction with system matrix derived from point source measurements

The quality of images reconstructed by statistical iterative methods depends on an accurate model of the relationship between image space and projection space through the system matrix The elements of the system matrix for the clinical Hi-Rez scanner were derived by processing the data measured for a point source at different positions in a portion of the field of view. These measured data included axial compression and azimuthal interleaving of adjacent projections. Measured data were corrected for crystal and geometrical efficiency. Then, a whole system matrix was derived by processing the responses in projection space. Such responses included both geometrical and detection physics components of the system matrix. The response was parameterized to correct for point source location and to smooth for projection noise. The model also accounts for axial compression (span) used on the scanner. The forward projector for iterative reconstruction was constructed using the estimated response parameters. This paper extends our previous work to fully three-dimensional. Experimental data were used to compare images reconstructed by the standard iterative reconstruction software and the one modeling the response function. The results showed that the modeling of the response function improves both spatial resolution and noise properties.     

Stereo calibration from correspondences of OTV projections

Stereo images have to be calibrated before stereo vision can recover three-dimensional information of the imaged scene. Position constraints over image point correspondences are traditionally used to solve the calibration problem. A method is described that uses angle constraints over correspondences of a particular type of image features, the projections of orthogonal trihedral vertices (OTV), for calibration. Computations of the rotation matrix and the translation vector are separable and the method has a closed-form solution. It also requires correspondences of only two vertex projections at minimum to recover all the transformation parameters which are recoverable from a stereo image pair. Extensive experimental results, including those on real images, are presented and they show that use of angle constraints is generally more accurate than position constraints alone     

A method for brain 3D surface reconstruction from MR images

Due to the encephalic tissues are highly irregular, three-dimensional （3D） modeling of brain always leads to compli- cated computing. In this paper, we explore an efficient method for brain surface reconstruction from magnetic reso- nance （MR） images of head, which is helpful to surgery planning and tumor localization. A heuristic algorithm is pro- posed foi＂ surface triangle mesh generation with preserved features, and the diagonal length is regarded as the heuristic information to optimize the shape of triangle. The experimental results show that our approach not only reduces the computational complexity, but also completes 3D visualization with good quality.     

Point-sampled 3D video of real-world scenes

This paper presents a point-sampled approach for capturing 3D video footage and subsequent re-rendering of real-world scenes. The acquisition system is composed of multiple sparsely placed 3D video bricks. The bricks contain a low-cost projector, two grayscale cameras and a high-resolution color camera. To improve on depth calculation we rely on structured light patterns. Texture images and pattern-augmented views of the scene are acquired simultaneously by time multiplexed projections of complementary patterns and synchronized camera exposures. High-resolution depth maps are extracted using depth-from-stereo algorithms performed on the acquired pattern images. The surface samples corresponding to the depth values are merged into a view-independent, point-based 3D data structure. This representation allows for efficient post-processing algorithms and leads to a high resulting rendering quality using enhanced probabilistic EWA volume splatting. In this paper, we focus on the 3D video acquisition system and necessary image and video processing techniques.     

Three-dimensional visualization of partially occluded objects using integral imaging

A three-dimensional (3-D) scene may contain foreground objects which partially occlude background objects. In this paper, we present experiments on reconstruction of the unobstructed view of the background objects from a partially occluded 3-D scene using an integral imaging system. We have reconstructed images of the scene at various distances of interest through computational and optical integral imaging reconstruction in order to obtain images of the occluding foreground objects and partially occluded background objects.