Representing 3D shapes based on implicit surface functions learned from RBF neural networks

We propose to represent the shape of 3D objects using a neural network classifier. The 3D shape is learned from a neural network, where Radial Basis Function (RBF) is applied as the activation function for each perceptron. The implicit functions derived from the neural network is a combination of radial basis functions, which can represent complex shapes. The use of RBF provides a rotation, translation and scaling invariant feature to represent the shape. We conduct experiments on a new prostate dataset and public datasets. Our testing results show that our neural network-based method can accurately represent various shapes.     

基于Unity3D的可变视点数立体图像合成算法

针对当前自由立体3D显示器的最佳观看距离固定和无法灵活调的等问题,本文在Unity3D平台上采用Cg语言编程,提出了可变视点数立体图像合成算法。该算法根据最佳观看距离与视点数之间的关系,调整光栅截距对应的像素点,实现了通过灵活改变视点数来改变自由立体3D显示的最佳观看距离。实验结果表明,通过此算法合成的立体图像可以调整显示的最佳观看距离,并且实验测得的视点数与计算所得的视点数误差较小。图像串扰较低、合成效率高,使观看者能在不同距离都能舒适观看立体图像。     

三维物体形状检测与重构系统的研究

三维物体形状检测与重构技术是计算机图像处理技术的一个分支,在众多领域有着广泛的应用前景。文中介绍了基于面结构光投影法的三维物体形状检测与重构系统,阐述了测量原理,建立了数学模型,给出了系统检测与重构的步骤,并根据检测得到的三维点云数据的特点,提出了适合本课题的简单、有效的三角网格化方法。实验结果证明该系统能够进行有效的检测与重构,具有很好的应用价值。     

基于形状匹配变形模型的三维人脸重构

为了从多幅人脸图像构造三维人脸结构,通常需要自动提取不同图像中的对应特征点,这往往是很难完成的.为了避免这个困难,本文建立了一个基于形状匹配的三维变形模型,在保证形状最佳匹配的条件下,实现对人脸图像姿态的估计和三维人脸重构.模型采用径向基函数对通用头部模型进行变形,用形状上下文来描述点之间的形状相似性,形状距离用来描述头部模型和人脸图像整体形状上的相似性,从而实现形状最佳匹配意义上的三维重构.实验表明,本文的算法只需要在人脸图像中提取特征点集,不需进行配准,就可以恢复出令人满意的三维头部结构.     

基于立体匹配技术的数字全息三维形貌重构

针对数字全息对物体三维形貌的重构与测量,提出了一种将数字全息术与立体匹配术相结合的三维测量方法。首先利用离轴菲涅耳数字全息系统,采集三维物体的单幅离轴菲涅耳全息图;然后将获取的数字全息图分为两个部分,分别进行数值再现,可以得到两个再现像,两个再现像存在视差。最后利用立体匹配算法获取两幅视图再现像的视差,根据几何关系获取物体的深度信息,重构物体的三维形貌。实验中,分别对不连续物体和连续物体进行三维形貌的重构,得到了准确的三维物体深度信息。数值模拟和初步实验结果表明该方法有效可行。     

基于单幅图片恢复人体三维形貌的研究

从单幅图片出发,通过简单的方法,根据人体的先验知识,从光学角度预测图片中的光源,并对人体图片进行2.5维恢复,以便以后从视频中恢复三维信息。实验表明,在一定条件下,将光源预测应用于人体表面三维形貌的复原是有成效的。本工作为从二维视频恢复2.5维信息提供了简单、实用、可改进的方法。     

一种基于MRF的单幅图像数据的三维重构方法研究

综合分析了常见的基于图像的三维重构方法的优缺点,提出了一种基于单张图像,采用马尔科夫随机场(MRF)推断3D位置和方向的3D重构方法。该算法首先将图片分割成多个小的区域(超像素块),并假定空间场景由许多很小的平面组成,超像素块与平面相互对应,对图像中每个超像素块求取出一组特征向量(纹理、颜色等),使用MRF模型化平面参数之间、超像素特征向量与平面参数之间的关系,采用监督学习的方式求取相关参数,求解MRF模型,并根据平面参数进行场景重建。这种算法不需对场景结构做明确的假定,因此较之以前的方法可以获得更多3D结构细节信息。用该方法对200张图片样本进行验算,样本中有60%实现了较为准确的3D重构。     

Direct method for rotation estimation from spherical images using 3D mesh surfaces with SPHARM representation

Since a large field of view obviously bears important advantages, the use of spherical images is becoming increasingly important in various computer vision and image processing applications. This paper presents a novel rotation estimation approach for spherical images based on 3D mesh representation of gray level intensity. Once the 3D meshes of the underlying spherical images are obtained, the 3D rotation can be estimated directly and efficiently, without feature extraction and matching process. Subsequently, we propose a direct method for 3D object rotation estimation using spherical harmonics representation with SVD decomposition and ICP algorithm for estimation refinement. Experimental results validate our approach and prove its suitability and robustness for rotation estimation. Moreover, it performs well against noisy images, brightness changes, image compression and occlusions. A comparative study of our proposed approach with four similar methods for 3D rotation estimation between spherical images, is realized to prove its effectiveness.     

激光自动变位测量汽车三维形状的研究

介绍了一种将激光和ＣＣＤ用于测量汽车等大型三维曲面形状的光学三角自动变位测量方法。分析了该测量方法的原理和过程。介绍了实验系统和测量实验，并给出了实测结果。较好地解决了测量精度与位移的关系。结果表明这种测量法具有一定的实用价值。     

基于数字散斑剪切干涉术的三维面型测试

基于数字散斑剪切干涉术的原理, 研究了物体面型的三维测试技术。使用倾斜反射镜的方法, 使物光束散斑场偏转,再由CCD记录偏转前和后物光束的散斑场。这种方法代替了传统的使用剪切元件空间上直接获得干涉条纹图, 减少了由于剪切元件分束引起的物光束的光能损失, 实现了数字散斑剪切干涉术测量物体面型。应用MATLAB软件编程, 控制剪切散斑干涉图的精确定向平移, 得到了4个干涉图, 实现了数字四步相移, 代替了传统的在参考光中利用位相控制器的机械相移, 从而实现了物体面型的三维测试。该方法不仅省去传统剪切散斑干涉术中需要对感光胶片进行化学湿处理的过程, 也省去了压电位移器等硬件。大量的实验表明, 该方法不仅简化了测试光学系统和实验的过程, 也提高了测试精度, 其精度可达到1/10波长。     

基于数字散斑剪切干涉术的三维面型测试

基于数字散斑剪切干涉术的原理,研究了物体面型的三维测试技术。使用倾斜反射镜的方法,使物光束散斑场偏转,再由CCD记录偏转前和后物光束的散斑场。这种方法代替了传统的使用剪切元件空间上直接获得干涉条纹图,减少了由于剪切元件分束引起的物光束的光能损失,实现了数字散斑剪切干涉术测量物体面型。应用MATLAB软件编程,控制剪切散斑干涉图的精确定向平移,得到了4个干涉图,实现了数字四步相移,代替了传统的在参考光中利用位相控制器的机械相移,从而实现了物体面型的三维测试。该方法不仅省去传统剪切散斑干涉术中需要对感光胶片进行化学湿处理的过程,也省去了压电位移器等硬件。大量的实验表明,该方法不仅简化了测试光学系统和实验的过程,也提高了测试精度,其精度可达到1/10波长。     

三维激光扫描测量系统的三维图像数据压缩

三维彩色激光扫描测量系统是利用计算机视觉的原理对物体进行三维测量和重建,以获取物体的三维图像。三维图像的数据量很大,文中研究了三维图像的压缩策略和算法,采用三角形条带的网格压缩方法,以及顶点数据的编码方法,可以把三维图像的存贮空间减小到压缩前数据的１／６,从而有效地解决了测量系统在实际应用环境下的图像存贮和传输问题。     

A Personalized Multifunctional 3D Printed Shape Memory-Displaying,Drug Releasing Tracheal Stent

The ability to engineer custom-made medical devices and to implant them minimally invasively are two important trends in modern surgery. The personalization of the device is achieved by 3D printing it, while the capacity to deploy it minimally invasively harnesses the shape memory behavior displayed by the inks used. This study introduces a 3D printed, shape memory-displaying tracheal stent based on novel, flexible photo-polymerizable inks comprising polypropylene glycol/polycaprolactone triblocks. This research introduces the in situ welding strategy, whereby thin and flexible layers of the stent are separately printed, sequentially deployed, and then welded together at the tracheal site. By doing so, the insertion profile of the device is dramatically reduced and its flexibility largely increased. Porous stents are 3D printed seeking to prevent mucus plugging. By combining more than one ink, their properties are further fine-tuned. Polyethylene glycol chains are covalently bonded to the stent surface to minimize biofilm formation, an important drawback of current tracheal stents. The in vitro cell viability and cell adhesion behavior of the treated surfaces reveal their compatibility and anti-adhesive behavior. In order to prevent implant-related infections, ciprofloxacin is added to the ink, and released in vitro over time, rendering the stent with antibacterial activity.     

狭小空间三维形貌测量方法的研究与实践

针对人的视线不易到达的狭小空间的物体的三维形貌测量在医疗诊断、工业设备的检测及故障诊断等领域有迫切的应用需求这一问题,设计了由数字微镜装置（DMD）芯片为核心的数字光条纹投影（DFP）单元、图像传导光纤、CCD摄像机以及计算机等构成的小型数字三维形貌测量系统。应用该系统对一直径为3cm的小球作了外形测量并进行了标定,进而在口腔内作了三维测量,并获得了令人满意的结果。据此可以将这一技术推广到航空、航天、工业检测、医疗诊断等领域。     

Tunable Superhydrophobicity from 3D Hierarchically Nano-Wrinkled Micro-Pyramidal Architectures

Interfacial surfaces with hierarchical structures have triggered intense research interest and been used in a broad range of applications ranging from anti-icing, anti-fogging, surface-enhanced Raman spectroscopy to catalytic reaction. However, the high-cost manufacturing processes and the limited durability remain as great challenges that need to be addressed, especially for superhydrophobicity. In this work, a novel hybrid approach for stretchable, transparent, and robust superhydrophobic surfaces is proposed, constructing micro-pyramid architectures with tunable hierarchical wrinkles efficiently and in an environmentally friendly manner. Due to the multiscale structures, excellent superhydrophobicity is obtained with contact angle of ≈172° and sliding angle ≈5° in a steady “Lotus” state. Further, the wear resistance and stability tests also suggest a superior performance under simulated severe real-world applications. Utilizing such multiscale synergistic co-operation effects can be an excellent manufacturing strategy and be extended to other surface engineering where hierarchical architectures are needed.     

4D Printed Shape Memory Anastomosis Ring with Controllable Shape Transformation and Degradation

Biofragmentable anastomosis ring (BAR) is an ideal sutureless alternative for intestinal connection that is frequently demanded in colonic surgery. However, it is challenging to insert a bulky BAR into the soft and slippery intestine. Here 4D printing of an anastomosis ring with shape memory capability is presented via fused deposition modeling (FDM) 3D printing. The shape memory anastomosis ring can recover from a compressed shape that facilitates the insertion to the permanent shape for connection and supporting. Degradation kinetics is tuned by controlling the blending composition of polylactic acid and poly(lactic-co-glycolic acid), so that the device can be excreted after the intestine healing. The shape recovery temperature is adjusted to 50 °C that the human body can withstand for a while. Grid structure and hook lock are designed and printed to guarantee dimension reduction upon programming and stable connection after shape recovery, respectively. A conceptual anastomotic operation shows the advantages and prospects of shape transformation. The 4D printing strategy may promote intestinal anastomosis development and inspire more opportunities for minimally invasive medical surgery.     

3D Printed Actuators: Reversibility,Relaxation, and Ratcheting

Additive manufacturing strives to combine any combination of materials into 3D functional structures and devices, ultimately opening up the possibility of 3D printed machines. It remains difficult to actuate such devices, thus limiting the scope of 3D printed machines to passive devices or necessitating the incorporation of external actuators that are manufactured differently. Here, 3D printed hybrid thermoplast/conducter bilayers are explored, which can be actuated by differential heating caused by externally controllable currents flowing through their conducting faces. The functionality of such actuators is uncovered and it is shown that they allow to 3D print, in one pass, simple flexible robotic structures that propel forward under step‐wise applied voltages. Moreover, exploiting the thermoplasticity of the nonconducting plastic parts at elevated temperatures, it is shown that how strong driving leads to irreversible deformations—a form of 4D printing—which also enlarges the range of linear response of the actuators. Finally, it is shown that how to leverage such thermoplastic relaxations to accumulate plastic deformations and obtain very large deformations by alternatively driving both layers of a bilayer; this is called ratcheting. The strategy is scalable and widely applicable, and opens up a new approach to reversible actuation and irreversible 4D printing of arbitrary structures and machines.     

Multi-view convolutional vision transformer for 3D object recognition

With the rapid development of three-dimensional (3D) vision technology and the increasing application of 3D objects, there is an urgent need for 3D object recognition in the fields of computer vision, virtual reality, and artificial intelligence robots. The view-based method projects 3D objects into two-dimensional (2D) images from different viewpoints and applies convolutional neural networks (CNN) to model the projected views. Although these methods have achieved excellent recognition performance, there is not sufficient information interaction between the features of different views in these methods. Inspired by the recent success achieved by vision transformer (ViT) in image recognition, we propose a hybrid network by taking advantage of CNN to extract multi-scale local information of each view, and of transformer to capture the relevance of multi-scale information between different views. To verify the effectiveness of our multi-view convolutional vision transformer (MVCVT), we conduct experiments on two public benchmarks, ModelNet40 and ModelNet10, and compare with those of some state-of-the-art methods. The final results show that MVCVT has competitive performance in 3D object recognition.     

板级立体组装凸点互联技术

通过对板级立体组装凸点互联焊盘设计技术的研究,找到了凸点焊盘设计的一种可靠的理论途径,很好地解决了凸点互联中对位精度控制的问题,实现了板级立体组装的凸点互联工艺技术。