边界模型布尔分割环自动识别

布尔分割环是设计特征在边界模型上造型的痕迹,其自动识别是后续设计特征自动重构的基础.在对特征造型过程形状演化分析基础上,提出一种布尔分割环几何表示和自动识别构建方法.首先依据边界的凹凸特性及顶点的邻接特性构建边界模型的特征顶点邻接图,并对邻接图开展顶点节点树搜索;然后将搜索方法分为初始节点选择、节点增长、搜索评估和反向传播4个步骤,在节点增长步骤中,采用相邻节点到特征面距离最小判定方法及顶点虚拟邻接构成伪边的方法,实现分割环的封闭和特征面的有效分割.实例结果表明,该方法能够从复杂特征相交形成的交线中自动识别和构造出布尔分割环.     

基于凸凹信号的网格分割

网格分割在网格参数化、纹理atlas图等几何处理问题中有着重要的应用,提出一种基于顶点或面凸凹信号的简单高效的网格分割算法,基于均匀支撑半径的顶点凸凹信号分析将顶点分为平坦点、凸点、凹点和特征点,先从平坦点进行平坦区域扩展,再从剩下的凸凹点出发进行凸凹区域扩展,最后根据顶点和边界边的光滑度进行区域竞争扩展;对于未能完全分割的简化程度高的模型,基于面的凸凹信号采用类似的过程进一步完成最后的分割,该算法可以快速地进行网格分割并能较好地保持网格特征,特别适用于CAD模型的分割。     

对称零空间准则下的LDA特征抽取方法

网格分割在网格参数化、纹理atlas图等几何处理问题中有着重要的应用.提出一种基于顶点或面凸凹信号的简单高效的网格分割算法.基于均匀支撑半径的顶点凸凹信号分析将顶点分为平坦点、凸点、凹点和特征点,先从平坦点进行平坦区域扩展,再从剩下的凸凹点出发进行凸凹区域扩展,最后根据顶点和边界边的光滑度进行区域竞争扩展;对于未能完全分割的简化程度高的模型,基于面的凸凹信号采用类似的过程进一步完成最后的分割.该算法可以快速地进行网格分割并能较好地保持网格特征,特别适用于CAD模型的分割.     

基于非线性扩散滤波的水平集模型MRI分割

基于曲线演化的图像分割模型在分割目标时需要在目标附近人为地构造一条曲线作为初始曲线,在此基础上进行演化得到目标边界.当初始曲线离目标边界较远时,影响模型分割的效率;当初始曲线离目标边界很近时,意味着需要过多的人为操作,这使得其时间效率较低且易出错.为此,在非线性扩散滤波的基础上,给出一种半自动初始曲线构造方法,该方法首先利用AOS算法对图像进行非线性扩散滤波,再利用区域信息快速地得到离目标边界很近的初始曲线.然后构造一种新的基于区域信息的速度函数,由水平集模型对其演化,得到了较好的结果.MRI分割实验表明了方法的有效性.     

一种新的基于区域竞争模型的水平集医学图像分割方法

传统的基于梯度模型的水平集分割方法在水平集曲线演化过程中存在着边界泄漏问题。针对这个问题,提出了一种基于改进区域竞争模型的水平集分割方法。本方法首先通过概率分布公式计算出水平集曲线属于目标区域和背景区域的概率;其次,将概率差值连同权重因子添加到水平集函数方程中,使曲线在演化过程中能量函数达到最小;最终,利用图像的区域信息提高水平集曲线识别边界的能力。实验结果表明该方法能够很好地实现医学图像的分割。     

一种新的基于背景的红外图像分割方法

针对红外图像中目标周围的背景往往比较均匀的特点,文中提出了一种基于背景区域的目标提取方法.首先利用局部熵信息检测到目标的位置,并在此基础上获得一个包含目标的感兴趣区域,最后通过基于直方图搜索的区域增长法来提取感兴趣区域中的背景部分,从而有效地将背景和目标进行了分割,实现了目标完整提取的目的.由于是在感兴趣区域进行提取而且对直方图进行搜索,克服了区域增长法计算量大的缺陷.试验结果证明,该方法能有效且完整地提取目标.     

基于局部特征的自适应快速图像分割模型

基于区域的活动轮廓模型如Chan-Vese(CV)模型等以其能较好的处理图像的模糊边界和复杂拓扑结构而广泛运用于图像分割中.然而基于灰度分布均匀假设,该模型对于含灰度不一致性的目标分割结果较差.此外,纹理是周期性重复出现的细节,依靠灰度信息无法正确检测.针对这些问题,提出一种基于局部特征的自适应快速图像分割模型.一方面,利用两种区域项检测卡通部分和纹理部分的特征信息,在自适应的局部块中提取局部统计信息以克服卡通部分的灰度不一致性;另一方面,利用自适应的局部块中的纹理特征来计算背景和目标区域的Kullback-Leibler (KL)距离以检测图像的纹理部分.进一步,基于分裂Bregman方法对该模型进行快速求解.分别对医学和纹理图像进行了实验,准确性和时效性都有显著提高.     

基于格式塔心理学原理的几何活动轮廓模型

基于格式塔心理学原理提出了一种几何活动轮廓模型,并将其应用于图像分割。当轮廓曲线远离目标边界时,应用格式塔心理学目标-背景原则,其能量函数主要由区域间差异性组成;当轮廓曲线位于目标边界附近时,应用格式塔心理学接近性原则,其能量函数主要由区域内一致性组成。该模型符合知觉特性,是几何活动轮廓模型的一般形式,且融合图像区域信息和边界信息。通过侧脑室和肿瘤医学图像分割实验,其结果表明,该模型对模糊边界图像的自动分割具有一定的普适性,能达到满意的分割效果。将该模型应用到多目标的免疫细胞图像分割中,能一次性完成将细胞质从细胞核和体液两种不同背景中分割出来的任务。     

基于先验知识和区域信息的Snake模型图像分割研究

提出基于先验知识和区域信息的参数活动轮廓模型(Snake模型)图像分割.用包含先验知识和区域信息的变力替换在气球力Snake模型中的恒定气球力,并应用于模糊边界图像分割.实验结果表明,该模型与初始轮廓曲线位置无关,能完成自动分割模糊边界图像的任务.另外,对于均值相等、方差不同的目标和背景两区域图像实行分割,该模型也能获得正确的分割结果.     

基于边界特征的三维模型分割

点云分割是三维模型检索、分类及重建的基础,为解决点云分割算法存在鲁棒性差、过分割和欠分割问题,提出一种基于边界特征的点云模型分割算法。将点云模型过分割为弱凸区域,利用巴氏距离判断相邻区域的相似性进行区域合并,采用改进的形状直径函数进行最终合并。由主流评价方法及实验证明,大多数模型可以取得良好的分割效果。     

基于Graph Cut与区域生长的连续CT图像分割算法

Graph Cut方法用于医学图像分割具有精度高,分割准确等优点,但处理每一幅图片都需要用户选定对象和背景,耗时较长.区域生长方法适于对面积不大的区域进行分割,分割速度快,但需要人工选取种子点,且在对比度低的情况下分割效果不理想.针对医学CT连续断层图像间相关性强特点,提出一种把Graph Cut方法和区域生长方法相结合的图像分割算法GCRGIS.首先使用Graph Cut法对连续断层图像的首幅图像进行分割,以分割出的图像轮廓作为后幅断层图像待生长区域的边缘,将边缘进行腐蚀后再进行区域生长,分割出目标图像.实验结果表明,该方法处理连续CT图像时仅需对首幅图像进行人工交互,在后续图像的分割中避免了每幅图像都要人工交互的繁琐,分割效果好,速度快.     

Constructing hexahedral shell meshes via volumetric polycube maps

Shells are three-dimensional structures. One dimension, the thickness, is much smaller than the other two dimensions. Shell structures can be widely found in many real-world objects. This paper presents a method to construct a layered hexahedral mesh for shell objects. Given a closed 2-manifold and the user-specified thickness, we construct the shell space using the distance field and then parameterize the shell space to a polycube domain. The volume parameterization induces the hexahedral tessellation in the object shell space. As a result, the constructed mesh is an all-hexahedral mesh in which most of the vertices are regular, i.e., the valence is 6 for interior vertices and 5 for boundary vertices. The mesh also has a layered structure, so that all layers have exactly the same tessellation. We prove that our parameterization is guaranteed to be bijective. As a result, the constructed hexahedral mesh is free of degeneracy, such as self-intersection, flip-over, etc. We also show that the iso-parametric line (in the thickness dimension) is orthogonal to the other two iso-parametric lines. We apply our algorithm to numerous real-world models of various geometry and topology. The promising experimental results demonstrate the efficacy of our algorithm. Although our main focus is to construct a hexahedral mesh by using volumetric polycube parameterization, the proposed framework is general that can be applied to other regular domains, such as cylinder and sphere, which is also demonstrated in the paper.     

Restoration of brick and stone relief from single rubbing images

We present a two-level approach for height map estimation from single images, aiming at restoring brick and stone relief(BSR) from their rubbing images in a visually plausible manner. In our approach, the base relief of the low frequency component is estimated automatically with a partial differential equation (PDE)-based mesh deformation scheme. A few vertices near the central area of the object region are selected and assigned with heights estimated by an erosion-based contour map. These vertices together with object boundary vertices, boundary normals as well as the partial differential properties of the mesh are taken as constraints to deform the mesh by minimizing a least-squares error functional. The high frequency detail is estimated directly from rubbing images automatically or optionally with minimal interactive processing. The final height map for a restored BSR is obtained by blending height maps of the base relief and high frequency detail. We demonstrate that our method can not only successfully restore several BSR maps from their rubbing images, but also restore some relief-like surfaces from photographic images.     

Artistic Low Poly rendering for images

This paper presents an automatic approach for generating low poly rendering of images, which is particularly popular in the recent art design community. Distinguishing from the traditional image triangulation methods for the sake of compression or vectorization, we propose some critical principles of such Low Poly rendering problem, and simulate the artists creation procedures straightforwardly. To produce the visual effects with clear boundaries, we constrain the vertices along the feature edges extracted from the input image. By employing the Voronoi diagram iteration guided by a feature flow field, the vertices in the result image well reflect the feature structure of the local shape. Moreover, with the salient region detection, we can achieve different mesh densities between the front object and the background. Some special color processing techniques are employed to make our result more artistic. Our method works well on a wide variety of images, no matter raster photographs or artificial images. Experiments show that our approach is able to generate satisfying results similar to the artwork created by professional artists.     

基于层次B样条的网格模型变形技术

针对常用的网格模型提出了一个基于层次B样条控制的三维网格空间变形框架．首先由用户交互地选取编辑区域,由程序完成编辑区域的参数化和均匀重采样;然后用层次B样条光顺拟合这些均匀采样点,所得的R样条曲面作为网格模型的基曲面,计算待编辑区域中网格顶点相对该B样条基曲面的局部坐标,该局部坐标平移、旋转不变,可视为模型的内蕴几何特征,并作为变形操作中的不变量．用户通过编辑层次B样条基曲面或者直接编辑三维网格模型,可实现多分辨率变形．实验结果表明：该方法操作直观方便,无论对整体还是局部变形,都能取得可控、可靠的变形效果,且采刚B样条曲面的控制手段易于与已有的造型系统合成．     

全局对比和背景先验驱动的显著目标检测

针对传统背景先验方法中背景提取不精确并且背景抑制能力弱的问题,提出了全局对比和背景先验驱动的显著目标检测方法。首先将图像分割为一系列感知均匀的超像素,再由全局颜色对比得到基于全局的显著图并计算得到前景种子点;然后将每个边界超像素与前景种子点做对比,筛选差异性较大的边界超像素作为背景种子点并计算得到基于背景的显著图;最后在融合基于全局和背景显著图的基础上,提出一种多兴趣点高斯模型的方法进一步抑制背景并整体高亮显著区域。在公开的MSRA-1000数据测试集上与6种主流方法进行对比实验,结果表明,所提出的显著性目标检测方法对复杂边界信息具有更强的鲁棒性,并能有效抑制背景噪声。     

一种保持纹理的网面简化算法

在计算机视觉领域,三维网面的简化不仅要求保持物体形状和拓扑关系,还要求保持物体表面法线,纹理,颜色和边缘等物体特征,以使计算机视觉系统能有效地表示,描述,识别和理解物体和场景,为此讨论了一种基于边操作（边收缩,边分裂）,并具有颜色或灰度纹理特征保持的三维网面的简化算法,该算法将网面不对称最大距离作为形状改变测度,将邻域内颜色或灰度最大改变量作为纹理改变测试,从而在大量简化模型数据的同时,有效地保持了模型的几何形状,拓扑关系,颜色或灰度特征,以及网面顶点均匀分布。     

网格模型的局部编辑算法

提出一种新的网格模型局部编辑算法，该算法可以精确地控制变形区域的大小、边界和变形点的位移，克服了FFD及其改进算法的缺点．首先交互地定义一个附着在模型表面的控制网格；然后建立模型变形区域与控制网格间点的映射，再依据变形要求来编辑控制网格；最后根据映射关系反算出模型变形区域点的新位置．控制网格可以是参数曲面的控制网格，也可采用一般三角网格或预先定义的网格模板．为达到精确变形的目的，对模型与控制网格重叠的区域进行自适应细分．该算法计算简便、易于实现，并能达到很好的效果．     

A label propagation method using spatial-spectral consistency for hyperspectral image classification

In this article, a label propagation approach with automatic seed selection is developed for hyperspectral image classification. The core idea is to combine pixel-wise classification results with spatial information described by a data graph. Using only the support vector machine (SVM) classifier on spectral features to tackle the hyperspectral classification task will produce results with a salt-and-pepper appearance. To overcome this limitation, the spatial information is incorporated by label propagation. The performance of label propagation is dependent on two points: the seeds and the connection graph. Generally, a limited number of labelled samples are available, which are considered as seeds in label propagation. However, the limited seeds will result in bad label propagation. Therefore, pseudo-seeds are automatically selected in local windows. Specifically, the pixels whose initial labels according to SVM are consistent with their most spatial neighbours are selected as seeds. Through seed selection, the number of seeds is greatly increased. Then, the label information of the selected seeds is propagated to their spatial neighbours using a data graph which is constructed according to the local structures in the image. Through seed selection and label propagation on the graph, the problem of salt-and-pepper appearance is solved elegantly – the noisy labels are highly suppressed and most of the structures are preserved. Competitive experimental results on a variety of hyperspectral data sets demonstrate the effectiveness of the proposed method.