A feature modeller for sculptured object modelling

There are numerous discussions concerning feature modelling. Most of these studies concentrate on regular shaped object modelling. The objective of this paper is to extend feature modelling coverage to include sculptured objects, so that feature technology is beneficial to modelling this class of objects. Feature modelling for a sculptured object consists of two basic elements: features and modelling operations. Feature semantics are declared by means of constraints in a feature class specification. These constraints are instantiated by the modelling operations during the modelling session. A feature modeller for sculptured objects is designed to generate the feature model from these constraints. A validation mechanism is introduced in the modeller to maintain the semantics of the feature model.     

用不变矩和边界方向进行形状检索

基于形状的图像检索一直以来是图像内容检索的一个难点问题，而目前采用周长、面积、边角率等描述形状的方法不能使形状检索达到理想的效果．本文提出了一种新的针对图像形状的检索方法．首先，用Canny算子对图像进行平滑处理，提取图像边界方向直方图特征、其次，用不变矩来描述图像形状的区域特征，不变矩特征不受图像的缩放、平移和旋转的影响．最后，为了克服不变矩只关心对象区域，而对图像边界忽视的缺点，提出了不变矩与边界方向特征相结合的方法，使得检索取得更好的效果．本文通过对医学图像的形状检索实验，给出了实验结果和结论．     

Feature-based ontological framework for semantic interoperability in product development

An essential requirement in integrating tasks in product development is to have a seamless exchange of product information through the entire product lifecycle. A key challenge in the integration is the exchange of shape semantics in terms of understandable labels and representations. A unified taxonomy is proposed to represent, classify, and extract shape features. This taxonomy is built using the Domain-Independent Form Feature (DIFF) model as the representation of features. All the shape features in a product model are classified under three main classes, namely, volumetric features, deformation features and free-form surface features. Shape feature ontology is developed using the unified taxonomy, which brings the shape features under a single reasoning framework. One-to-many reasoning framework is presented for mapping semantically equivalent information (label and representation) of the feature to be exchanged to target applications, and the reconstruction of the shape model automatically in that target application. An algorithm has been developed to extract the semantics of shape features and construct the model in the target application. The algorithm developed has been tested for shape models taken from literature and test cases are selected based on variations of topology and geometry. Results of exchanging product information are presented and discussed. Finally, the limitations of the proposed method for exchanging product information are explained.     

A cutting-tool-dependent approach for partitioning of sculptured surface

Topic of the paper is in the area of sculptured surface machining (SSM). The objectives of the paper are as follow. Firstly, to develop an approach that enables one detecting regions of sculptured surface those are not accessible for a cutting tool of a given design (i.e. regions, which cutter cannot reach without being obstructed by another portion of the part). Secondly, if we observe any not-machinable regions, the approach to be developed has enable one to subdivide the sculptured surface P onto the cutter-accessible and onto the cutter-not-accessible regions. To resolve the problem of sculptured surface partitioning, focal surfaces for the surfaces P and T were applied. Based on implementation of focal surfaces, cutting-tool-dependent characteristic surfaces are introduced. This enables derivation of equation for boundary curve, and enhancing the developed approach for locally extremal kinds of tangency of the surfaces P and T. The effectiveness of the proposed approach is verified from numerical simulation with the simple and clear examples. The main advantage of the developed approach over existing methods is that it incorporates topology not only of sculptured surface to be machined but also topology of the machining surface of a tool to be applied. This makes obtained results more fitted for engineering application. Taken as a whole, the topic covered in this paper enables one to develop reliable software for machining the sculptured surface on multi-axis CNC machine.     

A preprocess algorithm of filtering irrelevant information based on the minimum class difference

Whether a word (or a feature) should be included or excluded during the process of text classification could depend on a number of factors, such as the amount of information it represents, its appearance frequency and its meaning. The application context is another important factor that needs to be considered. A word may be able to represent the characteristic of a document in one application context but may not reflect its nature in another. This paper reports on an investigation into the selection of features for classification with the consideration of the application context of the documents to be processed. A new feature selection algorithm for text classification to be known as the PBMCD algorithm is proposed. This algorithm has been implemented and tested using three different data sets. The experiment results have shown that this algorithm cannot only filter out irrelevant features before the classification process but also can increase the classification accuracy. As a comparison, experiment results with other methods have also been presented.     

基于特征的离散网格模型表示与编辑技术

本文提出了一种基于特征的离散网格模型表示方法,能够表达传统三维网格模型中缺失的高层次信息,并以模型编辑为例显示了其应用价值.该特征结构利用特征线、特征面、特征组来建立离散网格的特征结构,用于描述模型的形状、约束、语义等信息,在原有网格模型的基础上构建了一个“超网格”.通过构建特征间的拓扑关系和约束关系,模型在编辑过程中能够保持特定的形状、结构属性,同时由于编辑操作造成的网格修改被限定在局部的特征区域内,从而提高了模型编辑的运算效率.     

一种多特征动态集成的3D模型检索方法

基于内容的3D模型检索是多媒体信息检索的热点研究问题之一,其主要的问题是提取3D模型的形状特征,但是采用单一特征很难保证检索系统对于任意输入查询模型都有很好的检索准确率。为了提高准确率,引入多种形状特征,通过加权求和在输出层融合,得到总体上模型间的相似度。每种特征采用动态权值,针对输入查询模型的不同,根据用户反馈对知识库中的权值自动更新。采用了形状分布特征和球面调和特征进行系统的建模,实验表明,文中方法比采用单特征检索的准确率大大提高。     

A Method for Integrating Modelling with Decomposition

１ＩｎｔｒｏｄｕｃｔｉｏｎＷｉｔｈｔｈｅｉｎｔｒｏｄｕｃｔｉｏｎｏｆｃｏｍｐｕｔｅｒｓｉｎｔｏｍａｉｎｓｔｒｅａｍｍａｎｕｆａｃｔｕｒｉｎｇａｃｔｉｖｉｔｉｅｓ，ｆｅａｔｕｒｅ－ｂａｓｅｄｇｅｏｍｅｔｒｉｃａｌｍｏｄｅｌｈａｓｂｅｃｏｍｅｔｈｅｋ...     

形状的全尺度可视化表示与识别

目的 视觉目标的形状特征表示和识别是图像领域中的重要问题。在实际应用中,视角、形变、遮挡和噪声等干扰因素造成识别精度较低,且大数据场景需要算法具有较高的学习效率。针对这些问题,本文提出一种全尺度可视化形状表示方法。方法 在尺度空间的所有尺度上对形状轮廓提取形状的不变量特征,获得形状的全尺度特征。将获得的全部特征紧凑地表示为单幅彩色图像,得到形状特征的可视化表示。将表示形状特征的彩色图像输入双路卷积网络模型,完成形状分类和检索任务。结果 通过对原始形状加入旋转、遮挡和噪声等不同干扰的定性实验,验证了本文方法具有旋转和缩放不变性,以及对铰接变换、遮挡和噪声等干扰的鲁棒性。在通用数据集上进行形状分类和形状检索的定量实验,所得准确率在不同数据集上均超过对比算法。在MPEG-7数据集上精度达到99.57%,对比算法的最好结果为98.84%。在铰接和射影变换数据集上皆达到100%的识别精度,而对比算法的最好结果分别为89.75%和95%。结论 本文提出的全尺度可视化形状表示方法,通过一幅彩色图像紧凑地表达了全部形状信息。通过卷积模型既学习了轮廓点间的形状特征关系,又学习了不同尺度间的形状特征关系。本文方法在视角变化、局部遮挡、铰接变形和噪声等干扰下能保持较高的识别正确率,可应用于图像采集干扰较多以及红外或深度图像的目标识别,并适用于大数据场景下的识别任务。     

基于形状特征与变形保持的动态模型简化

提出了一种基于形状特征与变形区域保持的动态表面多分辨率模型生成方法.该方法使用了基于形状特征的二次误差度量来计算边折叠代价,可以较好的保持模型表面特征.在计算整个变形动画中累加的边折叠代价时,加入相邻帧之间的变形程度信息,以保持变形程度较大区域的细节特征.最后基于整体的边折叠顺序,对每一帧模型进行细微的调整,以得到视觉失真最小的简化网格.文中方法的效率较高,易于实现,并且可以在变形网格的任意帧上生成高质量的、保持良好细节特征的简化模型.     

Local parameterization of freeform shapes using freeform feature recognition

Form feature modeling is a much used shape modeling technique that offers high-level control over a shape. When a feature-based interpretation of shape data is not available, e.g. when a shape is obtained by a laser range scanner or from a database of shapes, then the features must be reconstructed through feature recognition. Many methods for the recognition of machining features exist, but these methods cannot be used for freeform feature recognition, of which the complexity is much larger. In this paper, a new freeform feature recognition method is presented that is based on a new definition of the freeform feature concept. The method uses a three-step approach to feature recognition, in which first the global shape of a feature is matched to the target shape model. In a second step, this global shape is locally adapted to the target shape by adapting the definition of the feature. Finally, if the desired configuration of the feature has been determined, it can be used to reconstruct the target’s shape. In the first two steps, an evolutionary approach is taken to maximizing the similarity between the feature and the target shape. Finally, the target shape is reconstructed to incorporate the recognized feature. An extensive application example is given and the method is validated by applying it to a large number of artificially created test cases.     

A flexible and effective NC machining process reuse approach for similar subparts

NC machining process reuse is widely accepted as an effective strategy for engineers to generate the process plan with less time and lower cost. However, there has been very little research on how to reuse the NC machining process of similar subparts. As a result, most reusable NC machining process still has to remain in the repository as tacit knowledge, which can easily get lost due to oblivion. This paper proposes a novel NC machining process reuse approach for similar subparts in which existing NC machining process cases are described in association with machining features. First, a feature-based parameter-driven model is established to formalize the links between information imbedded in the machining feature and the parameters of cutting tools, drive geometries, and machining strategies. Then, the NC process parameter-driven characteristic of similar feature is revealed from the perspective of machining geometry, machining precision of the feature, and cutter geometry. Moreover, an NC process reusability assessment approach of similar pocket/subpart is presented using the pocket’s medial axis transform. Finally, the NC machining process inheritance mechanisms are explored to implement the NC machining process reuse automatically and efficiently. A prototype system based on CATIA has been developed to verify the effectiveness of the proposed approach.     

RGB-D语义分割：深度信息的选择使用

目的 在室内场景语义分割任务中,深度信息会在一定程度上提高分割精度。但是如何更有效地利用深度信息仍是一个开放性问题。当前方法大都引入全部深度信息,然而将全部深度信息和视觉特征组合在一起可能对模型产生干扰,原因是仅依靠视觉特征网络模型就能区分的不同物体,在引入深度信息后可能产生错误判断。此外,卷积核固有的几何结构限制了卷积神经网络的建模能力,可变形卷积(deformable convolution,DC)在一定程度上缓解了这个问题。但是可变形卷积中产生位置偏移的视觉特征空间深度信息相对不足,限制了进一步发展。基于上述问题,本文提出一种深度信息引导的特征提取(depth guided feature extraction,DFE)模块。方法 深度信息引导的特征提取模块包括深度信息引导的特征选择模块(depth guided feature selection,DFS)和深度信息嵌入的可变形卷积模块(depth embedded deformable convolution,DDC)。DFS可以筛选出关键的深度信息,自适应地调整深度信息引入视觉特征的比例,在网络模型需要时将深度信息嵌入视觉特征。DDC在额外深度信息的引入下,增强了可变形卷积的特征提取能力,可以根据物体形状提取更相关的特征。结果 为了验证方法的有效性,在NYUv2(New York University Depth Dataset V2)数据集上进行一系列消融实验并与当前最好的方法进行比较,使用平均交并比(mean intersection over union,mIoU)和平均像素准确率(pixel accuracy,PA)作为度量标准。结果显示,在NYUv2数据集上,本文方法的mIoU和PA分别为51.9%和77.6%,实现了较好的分割效果。结论 本文提出的深度信息引导的特征提取模块,可以自适应地调整深度信息嵌入视觉特征的程度,更加合理地利用深度信息,且在深度信息的作用下提高可变形卷积的特征提取能力。此外,本文提出的深度信息引导的特征提取模块可以比较方便地嵌入当下流行的特征提取网络中,提高网络的建模能力。     

A methodology for feature interaction detection in the AIN 0.1framework

We propose an integrated methodology for specifying AIN (advanced intelligent networks) and switch based features and analyzing their interactions in the AIN 0.1 framework. The specification of each individual feature is tied to the AIN call model and requires only a minimum amount of information in terms of control and data for interaction analysis. Once a feature is specified, its specification is then validated for consistency with respect to control and data. Interaction analysis is conducted for a set of features based on the sharing of call variables between the SSP and the SCP. With this approach, one can detect the following interactions involving AIN features: (1) side effects, where a call variable modified by one feature is used by another feature and (2) disabling, where one feature disconnects a call, preventing another feature from execution. We also develop a theory that is based on the computation of sequences of messages exchanged between the SSP and the SCP and their call variable usage. This theory is shown to dramatically reduce the number of cases considered during the analysis. A brief overview of a tool that makes use of this methodology to aid in the task of feature interaction detection is also given     

特征驱动的曲面重建

提出了基于散乱空间点集进行曲面重建的新方法,从点集的空间位置信息中提取待建曲面的内蕴特征量——法向和曲率,利用点集的这些特征信息来确定拓扑重建的搜索空间,采用面片生长的方式重建曲面。该方法在快速获得正确拓扑连接的同时,直接生成了用较少的面片就能保持曲面特征的优化网格。     

基于图像扩散速度模型和纹理信息的人脸活体检测

为了解决人脸身份认证中的欺诈问题,提出了一种基于图像扩散速度模型和纹理信息的人脸活体检测算法。真实人脸和虚假人脸图像的空间结构不同,为了提取这种差异特征,该方法使用各向异性扩散增强图像的边缘信息。然后,将原始图像与扩散后图像的差值作为图像的扩散速度,并构建扩散速度模型。接着使用局部二值算法提取图像扩散速度特征并训练分类器。真实人脸图像和虚假人脸图像之间存在很多差异特征,为了进一步提高人脸活体检测算法的泛化能力,该方法同时提取人脸图像的模糊程度特征和色彩纹理特征,通过特征矩阵级联的方法将两种特征进行融合,并训练另一个分类器。最后根据分类器输出概率加权融合的结果做出判决。实验结果表明,该算法能够快速有效地检测出虚假的人脸图像。     

3D clipping algorithm for feature mapping across domains

This paper describes an algorithm based on 3D clipping for mapping feature models across domains. The problem is motivated by the need to identify feature models corresponding to different domains. Feature mapping (also referred to as feature conversion) involves obtaining a feature model in one domain given a feature model in another. This is in contrast to feature extraction which works from the boundary representation of the part. Most techniques for feature mapping have focused on obtaining negative feature models only. We propose an algorithm that can convert a feature model with mixed features (both positive and negative) to a feature model containing either only positive or only negative features.The input to the algorithm is a feature model in one domain. The algorithm for mapping this model to another feature model is based on classification of faces of features in the model and 3D clipping. 3D clipping refers to the splitting of a solid by a surface. The feature mapping process involves three major steps. In the first step, faces forming the features in the input model are classified with respect to one another. The spatial arrangement of faces is used next to derive the dependency relationship amongst features in the input model and a Feature Relationship Graph (FRG) is constructed. In the second step, using the FRG, features are clustered and interactions between features (if any) are resolved. In the final step, the 3D clipping algorithm is used to determine the volumes corresponding to the features in the target domain. These volumes are then classified to identify the features for obtaining the feature model in the target domain. Multiple feature sets (where possible) can be obtained by varying the sequence of faces used for clipping. Results of implementation are presented.     

基于特征融合的图像情感语义分类

基于颜色或颜色-空间信息的图像分类方法,由于没有考虑图像中所含目标对象的形状特征,分类效果不够理想,以服装图像作为数据源,提出并设计了颜色-边缘方向角二维直方图,将图像的颜色特征与形状特征融合起来进行图像分类。图像中的低阶可视化特征与高阶情感概念之间有着密切的关联,分析了服装图像的颜色和形状的融合特征与情感之间的相关性,采用概率神经网络作为分类算法来完成情感语义分类,实验结果表明,该方法的分类精度有了明显的提高。     

阶层式三维形状环特征提取方法

针对已有的三维形状局部特征属性单一及缺乏空间结构信息的问题,提出了一种融合三维形状拓扑连接信息的阶层式特征提取框架,并得到具有平移不变性的三维形状环特征。首先,以三维形状底层特征提取为基础,进一步利用等测地线环的方式对特征点的局部区域进行建模,抽象出包含丰富空间几何结构信息的中层特征;然后,利用稀疏编码方式对中层特征进一步概括抽象,进而得到更具区分力和丰富信息的高层特征。将该高层特征与已有的尺度不变的热核描述子（SI-HKS）在三维形状对应和形状检索这两类任务中进行对比,该特征准确率分别提高了24.5个百分点和7.2个百分点。实验结果表明所提特征相较于已有的特征描述符具有更高的分辨率和识别度。