Search

X-ray computed tomography (CT) enables nondestructive evaluation and dimensional metrology using the reconstructed volumes of measured objects. Although the abundant applications of industrial X-ray CT exist, a high-quality reconstructed volume requires a long measurement time owing to sharp transmission images from dense views. Intensive X-rays with the exposure kept can be used to reduce the measurement time. However, the quality of the measured objects inevitably deteriorates as a result of the so-called penumbra effect caused by large focal spots in X-ray sources. This paper proposes rotational fine-tuning (RFT) for the acceleration of CT scanning with the quality kept. First, sharp transmission images from sparse views are obtained in appropriate imaging conditions. Subsequently, blurry transmission images from dense views are obtained using a larger focal spot size and a shorter exposure time. The acquired blurry images are deblurred by convolutional neural networks fine-tuned using several pairs of sharp and blurry images obtained at the corresponding projection angles, and linear interpolation integrates the deblurred images to generate the final RFT output. The proposed method indirectly reduces the measurement time because no lengthy acquisition time is required, and fine-tuning and blurring are rapid when using GPUs. In the experiments, the comparison of PSNRs between blurry and deblurred reconstructed volumes is shown. To reveal the effect of the proposed method on dimensional metrology, the improvement in the surface deviations of the stepped cylinder is also shown. Finally, the improvement of the pore spaces in a porous aluminum is shown to assess the aspect of nondestructive evaluation. Overall, experimental results demonstrate that the proposed method can perform the acceleration of X-ray CT scanning with the quality kept.

Using a 3D scanner to inspect a 3D print is becoming more significant as the spectrum of applications for fused filament fabrication expands. We use the tool path, which describes the 3D printing p...

Making arts and crafts is an essential application of 3D printing. However, typically, 3D printers have limited resolution; thus, the perceptual quality of the result is always low, mainly when the input mesh is a relief. To address this problem using existing 3D printing technology, we only operate the shape of the input triangle mesh. To improve the perceptual quality of a 3D printed product, we propose an integrated mesh processing that comprises feature extraction, 3D print preview, feature preservation test, and shape enhancement. The proposed method can identify and enlarge features that need to be enhanced without large-scale deformation. Besides, to improve ease of use, intermediate processes are visualized via user interfaces. To evaluate the proposed method, the processed triangle meshes are 3D printed. The effectiveness of the proposed approach is confirmed by comparing photographs of the original 3D prints and the enhanced 3D prints.

All terrain cranes often work in construction sites. Blind spots, limited information, and high mental workload are problems encountered by crane operators. A top-view camera mounted on the boom head offers a valuable perspective on the workspace that can help eliminate blind spots and provide the basis for assisting operation. In this study, a visual 2D map of a crane workspace is generated from images captured by a top-view camera. Various types of information can be overlaid on this visual 2D map to assist the operator, such as recording the operation and projecting the boom head’s expected path through the workspace. Herein, the process of generating a visual map by stitching and locating the boom head trajectory in that visual map is described. Preliminary proof-of-concept tests show that a precise map and projected trajectories can be generated via image-processing techniques that discriminate foreground objects from the scene below the crane. The location error is analyzed and verified to confirm its applicability. These results show a way to help the operator make more precise operation easily and reduce the operator’s mental workload.

Good visualizations of computational fluid dynamics (CFD) results are key to making the best use of them. In many cases, CFD results involve multidimensional data embedded in 3D, and it can be difficult to recognize such complicated 3D shapes when they are projected onto 2D displays. A promising alternative approach uses a 3D printer to build real objects representing the CFD results. In this paper, we present a method of 3D printing the isosurfaces of CFD results to visualize them. However, there is a problem that the isosurfaces are not connected. Therefore, to prevent the structure from breaking up, we must find a way to connect them, and we achieve this by using beams to generate a support structure. We take a structural mechanics approach, using an algorithm to optimize the support structure such that the printed object does not break under gravity, enabling us to hold it safely for observation.

Eye tracking technology is widely applied to detect user’s attention in a 2D field, such as web page design, package design, and shooting games. However, because our surroundings primarily consist of 3D objects, applications will be expanded if there is an effective method to obtain and display user’s 3D gaze fixation. In this research, a methodology is proposed to demonstrate the user’s 3D gaze fixation on a digital model of a scene using only a pair of eye tracking glasses. The eye tracking glasses record user’s gaze data and scene video. Thus, using image-based 3D reconstruction, a 3D model of the scene can be reconstructed from the frame images; simultaneously, the transformation matrix of each frame image can be evaluated to find 3D gaze fixation on the 3D model. In addition, a method that demonstrates multiple users’ 3D gaze fixation on the same digital model is presented to analyze gaze distinction between different subjects. With this preliminary development, this approach shows potential to be applied to a larger environment and conduct a more reliable investigation.

The analysis of existing objects for developing new products is called reverse engineering and is a common process in industrial manufacturing. X-ray CT scanning is a powerful reverse-engineering tool that can acquire the entire geometry of an assembled object, including its inner structure, without disassembly. Before X-ray CT scanning can be used to accurately extract the parts of an assembled object from its CT volume, several challenges remain: CT artifacts, the non-uniqueness of the CT values, and limited resolution. Currently, an operator usually manually segments the target parts of the CT volume at voxel-level accuracy, which takes a long time. Hence, this paper proposes SegMo, a segmentation system which can perform segmentation of an assembled object to sub-voxel accuracy more quickly and with less effort than conventional methods. These advantages are achieved by suggesting the part boundary candidates that are coded as a hierarchical decomposition of the CT volume based on a Morse complex and polygonization at sub-voxel accuracy. SegMo consists of three steps: the generation of a hierarchical decomposition of a CT volume, segmentation using this hierarchy, and polygonization of the segmentation result. In this paper, we demonstrate how SegMo can achieve accurate and efficient segmentation on CT volumes of several complex assembled objects.

X-ray Computer Tomography systems (XCT) which can visualize internal structures of objects are very useful tool for detecting defects of parts. Their most common application is inspection of parts made by metal casting or by resin molding. These parts often contain voids, which are small empty regions inside the parts. XCT can scan these parts to generate the 3D CT images of their internal structure involving voids. The parts usually include both unacceptable and acceptable voids. What is practically needed for inspection is to detect unacceptable voids. In this paper we propose our unique classification method based on a concept so called golden part. We compute a golden part by averaging images of the parts containing only acceptable voids. Then the inspection of a test part can be made by comparing the CT image of the test part and the golden part. If the voids of the test part exist only in the void region of the golden part, the test part is passed while if some voids exist elsewhere, it is failed. We implemented a prototype inspection system and apply it to a test case. Its examples demonstrate the usefulness of our method.

The teeth segmentation plays a key role in the orthodontics analysis, but, due to their irregularity, it is a complex task. Irregular surfaces segmentation is still an open issue. Whereas some state-of-art algorithms are either very time consuming or not very precise, the region growth algorithm performs well in both aspects. In this project, the region growth algorithm was adapted to reduce noise and improve the performance with a pre-processing step. In this step, the curvature was analysed in order to obtain the boundaries and apply the region growth. Two method were developed: a manual algorithm and an automatic application. The manual algorithm was capable of segmenting all teeth correctly in the experiments. However, the automatic variant did not achieve the same results. Moreover, some holes must be removed in order to perform simulation on the results.

Good assembly instructions improve the effectiveness of assembly operations and training. The development of augmented reality (AR) has enhanced the experience of assembly operations by showing computer-aided design (CAD) models at installation locations to an operator wearing an AR head-mounted display (HMD). In such operations, it is important to detect whether misassembly occurs. This research proposes two methods to construct an AR assembly assistance system using only an HMD of Microsoft HoloLens. One method is coordinate calibration to make the CAD models display correctly at desired locations by aligning a reference CAD model to the corresponding real object’s position. The other evaluates in real time whether misalignment has occurred by comparing the depth images of the CAD model and the real object for efficiency. With the preliminary development, we can use only an HMD to achieve the AR-based assembly instruction and verification to some extent.

Adjective-based design is a method that translates human perception into design parameters quantitatively in order to achieve better understanding between designers and clients. In this approach, adjectives are used to describe product designs, which are generated via design parameters in terms of geometry. As a requirement of the concept, relations between hull adjectives (e.g., comfortable and aesthetic) and design parameters (e.g., length and width) are learned via a machine-learning algorithm. Nevertheless, the relations cannot be represented by some of the design parameters, although they are in the learning process. This issue shows that the parameters do not impact the adjective choices but add noises to the learning process. Therefore, in this study, visual evaluations are made using eye tracking technology for screening the parameters based on their attractiveness and establishing relations between the attractive ones and the adjectives to enhance quality of the relation representations. Eye tracking is used in perceptual research, which proves the existence of correlations between gaze data and human preferences. The main advantage of eye tracking is that reliable human perception data can more likely be collected compared to the user tests, since the evaluation is based on subjects’ attention rather than applying solely questionnaires that are limited by the question content. In light of the benefits and finding, an eye tracking device is used to collect gaze data, and then, eye tracking tools such as Area of Interest (AOI), scan path, and heat map are used to evaluate attractiveness of the design parameters. Finally, regression analysis is used to represent relations between gaze data of design parameters and the adjectives.

Eye tracking is a technology that has quickly become a commonplace tool for evaluating package and webpage design. In such design processes, static two-dimensional images are shown on a computer screen while a subject's gaze where he or she looks is measured via an eye tracking device. The collected gaze fixation data are then visualized and analyzed via gaze plots and heat maps. Such evaluations using two-dimensional images are often too limited to analyze gaze on three-dimensional physical objects such as products because users look at them not from a single point of view but rather from various angles. Therefore in this study we propose methods for collecting gaze fixation data for a three-dimensional model of a given product and visualizing corresponding gaze plots and heat maps also in three dimensions. To achieve our goals, we used a wearable eye-tracking device, i.e., eye-tracking glasses. Further, we implemented a prototype system to demonstrate its advantages in comparison with two-dimensional gaze fixation methods. Highlights Proposing a method for collecting gaze fixation data for a three-dimensional model of a given product. Proposing two visualization methods for three dimensional gaze data; gaze plots and heat maps. Proposed system was applied to two practical examples of hair dryer and car interior.

In this paper, a generative design approach is proposed that involves the users’ psychological aspect in the design space exploration stage to create distinct design alternatives. Users’ perceptual judgment about designs is extracted as a psycho-physical distance metric, which is then integrated into the design exploration step to generate design alternatives for the parametric computer-aided design (CAD) shapes. To do this, a CAD model is first parametrized by defining geometric parameters and determining ranges of these parameters. Initial design alternatives for the CAD model are generated using Euclidean distance-based sampling teaching–learning-based optimization (S-TLBO), which is recently proposed and can sample N space-filling design alternatives in the design space. Similar designs are then clustered, and a user study is conducted to capture the subjects’ perceptual response for the dissimilarities between the cluster pairs. In addition, a furthest-point-sorting technique is introduced to equalize the number of designs in the clusters, which are being compared by the subjects in the user study. Afterward, nonlinear regression analyses are carried out to construct a mathematical correlation between the subjects’ perceptual response and geometric parameters in the form of a psycho-physical distance metric. Finally, a psycho-physical distance metric obtained is utilized to explore distinct design alternatives for the CAD model. Another user study is designed to compare the diversification between the designs when the Euclidean and the suggested psycho-physical distance metrics are utilized. According to the user study, designs generated with the latter metric are more distinct.

Ceramic matrix composite (CMC) is a material with high thermostability. Since it is lower in weight than metals realizing the same thermostability, it has been attracting increasing attention in many fields. It has an inner fabric structure made of ceramics (SiC), and the yarns of the fabric give this material rather high stiffness in the directions the yarns run. To guarantee the stiffness of the material, it is necessary to inspect the yarns. X-ray CT scanning, a non-destructive inspection technique, is one of the best ways to do this. However, the quality of a CT volume of SiC fabric tends to be very low, and the resolution is generally also low because of the restriction on the time given for the inspection and the relatively large size of CMC parts. This paper presents an algorithm for computing the directions of the yarns in an SiC fabric from a low quality CT volume, and it proposes a way to visualize the computed directions for a better recognition of the directions. It also presents some experimental results that show the effects of the proposed algorithms.

A generative design system is proposed to produce distinct design variations from an example design supplied by users. The designs are represented using profile curves and generated under geometric rules. This system is useful for product aesthetic design, which is crucial for the global market that requires creativity under limited conditions such as manufacturing methods, cost, durability and so on. Since aesthetics must be combined with such conditions, designers must create many design alternatives and select the ones, which satisfy all defined requirements. Therefore, the conceptual design stage may take long time specifically when creative ideas are needed. The proposed approach can generate the design alternatives by minimum effort thanks to our easy to use system, where users only need to define points of the product profile by simply mouse-clicking on the product image. Once the points are supplied, the product profile curve is automatically defined using a composite cubic Bezier curve. It is modified to generate design variation under geometric constraints defined by geometric shapes such as circles, triangles and ellipses. Similarity measures, anisotropy ratio of the transformation of generated samples to the example profile and modified Hausdorff distance (MHD) between the sample profiles, are used to constrain generating irrelevant samples to the example profile, but ensuring the generated samples are different enough from each other and from the example profile. All these constraints are used in the proposed search algorithm to explore the samples.

2.5D printing system, Mofrel, is a new technology developed by Casio Computer Co. Ltd. in Japan. Roughly describing, it is an ink-jet color printer with a unique capability to generate swelling on the printer sheet. As the height of the swelling can be controlled, this technology is called 2.5D printing. It requires special printer sheet of a three layer structure. The height of swells over the sheet is controlled by a grey scale image printed on the back side of the sheet. Blacker regions produce higher swells while whiter regions produce lower swells. This grey scale ‘heating patterns’ are not generally available and manually generated by experts. In this study we propose a method to automatically generate heating patterns for 2.5D printing of topographic contour maps. We first generate a height field corresponding to the input contour map then convert it to the grey scale image to be used as a heating pattern. Our method is based on well-known Poisson image editing approach with which the height field can be computed by solving the Poisson equation with a gradient field of the height field. We propose three different estimation methods of the gradient field. The computed height field is then converted to a heating pattern. We developed a prototype software based on this approach and conducted experiments to evaluate the computed height fields and those of the actual printed sheet.

Inspired by computed tomography (CT), this paper presents a novel surface reconstruction algorithm, tomographic surface reconstruction, to reconstruct a surface mesh from a point cloud equipped with oriented normals. In the process of scanning a real object using an X-ray CT system, it generates a sinogram consisting of projection images that are maps of X-ray transmission lengths, and then, a tomogram (CT volume) is reconstructed from the sinogram. A hole-free surface mesh is then easily obtained by polygonizing an isosurface. To adopt this CT paradigm to surface reconstruction from a point cloud, only a scheme to generate a sinogram from a point cloud is required. The value of a sinogram for surface reconstruction can be defined as the sum of the distances between the intersecting points of a ray and the underlying surface, which are defined as the maxima of the point density. While ordinary CT scanning uses projection directions which share a single rotation axis, tomographic surface reconstruction adopts randomly selected projection directions and successfully improved the reconstruction robustness. By applying an iterative CT reconstruction to the sinogram, the algorithm generates a tomogram whose boundary between the foreground and background approximates the surface of the object. The effectiveness for a point cloud with a lack of sampling and outliers is demonstrated from experimental results. Graphical abstractDisplay Omitted HighlightsA novel approach for surface reconstruction using a CT-reconstruction algorithm.Enhancing robustness to compute transmission-length for a point cloud.Robustness over noise and outliers is demonstrated.

Many types of artifacts appear in X-ray computed tomography (CT) volume data, which influence measurement quality of industrial cone beam X-ray CT. Most of those artifacts are associated to CT scanning parameters; therefore, a good scanning parameter setting can weaken the influence to improve measurement accuracy. This paper presents a simulation method for evaluating CT scanning parameters for dimensional metrology. The method can aid CT metrology to achieve high measurement accuracy. In the method, image entropy is used as a criterion to evaluate the quality of CT volume data. For entropy calculation of CT volume data, a detailed description about bin width and entropy zone is given. The relationship between entropy values of CT volume data and error parameters of CT metrology is shown and discussed. By use of this method, mainly we focus on specimen orientation evaluation, and some other typical scanning parameters are used to evaluate the proposed method. Two typical specimens are used to evaluate the performance of the proposed method.

A prediction model is used to predict subjective responses of crane operators with respect to different designs of In-Vehicle Visual Support (IVVS). Selected gaze metrics are used as objective metrics to minimize prejudice, which is commonly caused by subjective measures. Experiments are carried out using crane simulator to measure glance behavior of novice operators and the 3D perspective projection method is used for autonomous mapping of gaze fixations to dynamic Area-of-Interests (AOIs). Subjective responses, such as operators' emotion and usability of IVVS, are evaluated using the Likert scale of the Semantic Differential method. Correlation between gaze metrics and subjective responses is established using multiple linear regression. Glance behavior exhibits a statistically significant difference when information on IVVS is perceived as useful to ease operation and reduce tension. Despite this, there are no significant signs of distraction. Glance behavior is found to be a reliable sub-conscious indicator of subjective response and distraction. More importantly, the proposed gaze metrics are found to be a good representation of glance behavior, such as randomness and distribution of attention. The methods and findings are useful to evaluate impact of IVVS, which is becoming more common in many other applications.

This study described a method of using glance behavior as design indices of In-Vehicle Visual Support (IVVS) system. The method was implemented to evaluate information content of the IVVS to facilitate operation of a mobile crane, which was challenging due to depth perception and load oscillation. The results suggested information content was well-discriminated using gaze behavior, which indicated its feasibility as design indices. Most importantly, emotional response was well-defined using gaze behavior. By not requiring self-evaluation of emotional response after the experimental procedure, design of IVVS could be evaluated by observing gaze behavior simultaneously during the experiment. In addition, elements of bias and social masking could be minimized because gaze response is sub-conscious.

Recently, to meet the requirement of machine industry, there has been an increased focus on the development of porous metal as a high-strength material despite its low density. To evaluate its properties and quality, material testing is usually conducted. However, it is more efficient to perform computer simulation evaluations using finite element analysis. The X-ray computed-tomography scanning technique enables us to obtain the information regarding the internal structure of the metal. Furthermore, a reconstruction algorithm produces volume data of the test object. In general, conventional methods are utilized to generate mesh data from volume data for finite element analysis, but a key drawback is that they generate too many elements, resulting in high computational cost. We propose an approach to generate meshes for porous structures by modeling each pore using spheres from volume data.

X-ray CT scanning is recently becoming a popular technology for industrial applications as well as medical ones. Since the geometrical accuracy is often important for industrial applications, more precise methods for processing CT volumes are required. This paper proposes a method for extracting the multi-material interfaces on CT volumes obtained by industrial CT scanners. Instead of extracting isosurfaces we detect edge-points, at which the norm of the volume gradient takes a local maximum in the gradient direction, and then interpolate the points as the material interfaces represented by the zero-level of compounded implicit functions. In order to achieve a robust material-identification, multilabel graph-cut is utilized in our method. Using edge-points, we can reduce the inaccuracy caused by CT scanning artifacts.

Three-dimensional geometric morphometric techniques have been widely used in quantitative comparisons of craniofacial morphology in humans and nonhuman primates. However, few anatomical landmarks can actually be defined on the neurocranium. In this study, an alternative method is proposed for defining semi-landmarks on neurocranial surfaces for use in detailed analysis of cranial shape. Specifically, midsagittal, nuchal, and temporal lines were approximated using Bezier curves and equally spaced points along each of the curves were defined as semi-landmarks. The shortest paths connecting pairs of anatomical landmarks as well as semi-landmarks were then calculated in order to represent the surface morphology between landmarks using equally spaced points along the paths. To evaluate the efficacy of this method, the previously outlined technique was used in morphological analysis of sexual dimorphism in modern Japanese crania. The study sample comprised 22 specimens that were used to generate 110 anatomical semi-landmarks, which were used in geometric morphometric analysis. Although variations due to sexual dimorphism in human crania are very small, differences could be identified using the proposed landmark placement, which demonstrated the efficacy of the proposed method. Am J Phys Anthropol 151:658–666, 2013. © 2013 Wiley Periodicals, Inc.

The authors present a method for extracting polygon data of endocranial surfaces from CT images of human crania. Based on the fact that the endocast is the largest empty space in the crania, we automate a procedure for endocast extraction by integrating several image processing techniques. Given CT images of human crania, the proposed method extracts endocranial surfaces by the following three steps. The first step is binarization in order to fill void structures, such as diploic space and cracks in the skull. We use a void detection method based on mathematical morphology. The second step is watershed-based segmentation of the endocranial part from the binary image of the CT image. Here, we introduce an automatic initial seed assignment method for the endocranial region using the distance field of the binary image. The final step is partial polygonization of the CT images using the segmentation results as mask images. The resulting polygons represent only the endocranial part, and the closed manifold surfaces are computed even though the endocast is not isolated in the cranium. Since only the isovalue threshold and the size of void structures are required, the procedure is not dependent on the experience of the user. The present paper also demonstrates that the proposed method can extract polygon data of endocasts from CT images of various crania.

This paper proposes skill metrics that analyzes the gaze fixation pattern of mobile crane operators. This study focuses on the fixation data because they commonly represent visual information processing. First, scenes of crane operation are divided into content-based Area of Interests (AOIs) and the Markov Chains is used to model the gaze transitions between these AOIs. Four metrics were introduced to interpret the model at different expertise levels. Results suggest that experienced operators exhibit lower metrics compared to their novice counterparts, and adapted entropy measures exhibit similar patterns as the original ones in the previous study. Most importantly, the proposed metrics are able to address the following issues, i.e. large and sparse model, as highlighted in the previous study.

Vision constitutes a significant part of information input for Human Machine Interaction (HMI), and the understanding of gaze characteristics such as stability, focus, and duration is promising to design a good User Interface (UI). This paper defines HMI events as UI design factors and identifies its correlation with users’ feedback, i.e. Gaze Metrics (GM) and affect. The observation of the trilateral relationship between these parameters during a pilot testing offers insights to designers on how to improve UI design to enhance usability and attractiveness.

We present a framework for reuse of digital mock-up models. Our idea is to find reusable models in the archives by shape matching. Since the reusable models are already designed by operators in the past projects, we can easily create various types of data by using existing models. Indeed, we demonstrate this framework is efficient for surface simplification and CAE mesh generation problems.

We propose geometric quality indicators for evaluating the quality of point clouds (sets of scanned points). These indicators represent aspects of quality often considered in practical scanning procedures, such as cloud thickness and cloud density. We defined the indicators mathematically and developed software to compute them, then conducted experiments to evaluate the indicators for point clouds obtained by scanning the same samples with different types of surface scanners and scanning procedures. The results showed that the indicators are capable of highlighting various aspects of point cloud quality.

W e developed an interactive system to design a customized cover for a given three-dimensional (3D) object such as a camera, teapot, or car. The system first computes the convex hull of the input geometry. The user segments it into several cloth patches by drawing on the 3D surface. This paper provides two technical contributions. First, it introduces a specialized flattening algorithm for cover patches. It makes each two-dimensional edge in the flattened pattern equal to or longer than the original 3D edge; a smaller patch would fail to cover the object, and a larger patch would result in extra wrinkles. Second, it introduces a mechanism to verify that the user-specified opening would be large enough for the object to be removed. Starting with the initial configuration, the system virtually "pulls" the object out of the cover while avoiding excessive stretching of cloth patches. We used the system to design real covers and confirmed that it functions as intended.

A knitted animal is made of a closed surface consisting of several knitted patches knitted out of yarn and stuffed with cotton (Fig. 1). We introduce a system to create a knitting pattern from a given 3D surface model (mainly designed for rotund animal models). A knitting pattern is an instructional diagram describing how to knit yarn to obtain a desired shape. Since the creation of knitting patterns requires special skill, this is difficult for nonprofessionals. Our system automates the process and allows anyone to obtain his or her original knitting patterns from a 3D model. The system first covers the surface of the model with parallel winding strips of constant width. The system then samples the strip at constant intervals to convert it into a knitting pattern. The result is presented in a standard visual format so that the user can easily refer it during actual knitting. We show several examples of knitted animals created using the system.

Summary form only given. X-ray CT (computed tomography) has experienced tremendous growth in industrial application in recent years, and acquiring information about the mechanical parts from CT data has been a great challenge for researchers. This paper presents a new method for the registration of a CT volumetric model of an assembly of parts with a CAD mesh model of a part of the assembly using ICP (iterative closest point) registration method. A few steps to extract feature points should be done as a preprocessing step of the volumetric model and the mesh model before applying the ICP registration algorithm, since the volumetric model and the mesh model are different in their data representation. This preprocessing step is important in order to unify the input of the ICP algorithm, and contributes to the robustness and the speed of the registration process.

X-ray computed tomography (CT) can nondestructively inspect an object and can clearly, accurately, and intuitively display its internal structure, composition, texture, and damage. In industry this technology was initially used for material analysis and nondestructive testing and evaluation. Recently, as an alternative to optical and tactile measurement devices, CT has entered industrial use for dimensional metrology. Unfortunately, industrial-level accuracy is very difficult to attain with CT for various reasons. In this paper we analyze one of the most serious effects, the Feldkamp–Davis–Kress (FDK) effect, which can be observed in most of the common X-ray CT scanners with a cone beam. The FDK is the reconstruction algorithm widely accepted as a standard reconstruction method for cone-beam type of CT because of its computation efficiency. However, this algorithm merely provides an approximate result. An accurate measurement result can be obtained only in the case of small cone angle. We aim at analyzing the FDK effect independently from other kinds of artifacts. In a practical CT scanning situation, various kinds of artifacts appear in the reconstruction results; thus, we apply a simulation to obtain projection images without noise (scattering, beam hardening, etc.). Then, the FDK algorithm is applied to these projection images to reconstruct CT images so that only the FDK effect can be observed in the reconstructed CT images. Based on this approach, we conducted quantitative analysis on the FDK effect using numerical phantoms of the sphere and stepped cylinders that may be adopted as ISO reference standards for dimensional metrology using X-ray CT scanners. This paper describes the evaluation workflow and discusses the cause of the FDK effect on the measurement of the sphere and the stepped cylinders. Particular attention is given to the evaluation of the error distribution feature on different spatial positions. After discussing the error feature, a method for improving measurement accuracy is proposed.

Automatic creation of B-rep models of engineering objects from freehand sketches would benefit designers. One step aims to take a line drawing (with hidden lines removed), and from it deduce an initial three-dimensional (3D) geometric realization of the visible part of the object, including junction and line labels, and depth coordinates. Most methods for producing this frontal geometry use line labeling, which takes little or no account of geometry. Thus, the line labels produced can be unreliable. Our alternative approach inflates a drawing to produce provisional depth coordinates, and from these makes deductions about line labels. Assuming many edges in the drawing are parallel to one of three main orthogonal directions, we first attempt to identify groups of parallel lines aligned with the three major axes of the object. From these, we create and solve a linear system of equations relating vertex coordinates, in the coordinate system of the major axes. We then inflate the drawing in a coordinate system based on the plane of the drawing and depth perpendicular to it. Finally, we use this geometry to identify which lines in the drawing correspond to convex, concave, or occluding edges. We discuss alternative realizations of some of the concepts, how to cope with nonisometric-projection drawings, and how to combine this approach with other labeling techniques to gain the benefits of each. We test our approach using sample drawings chosen to be representative of engineering objects. These highlight difficulties often overlooked in previous papers on line labeling. Our new approach has significant benefits.

本稿では多面体の新しい展開図生成手法を提案する.本手法では, はじめに全ての面を同一平面上に離散的に配置し, これらの面を3次元空間での面の接続情報を元に徐々に集約していくことで展開図を構築する.本稿ではこのような展開手法を「集約法」, 従来の手法を「逐次法」と呼ぶことととし, 両者の手法によって生成される展開図の比較を行う.また, 新しく提案する集約法を用いることで, 3次元空間での面の向きを展開図上でも維持することや, 左右対称性を維持することが容易となる.これにより, 組立て方を把握しやすい展開図の生成を行えることを示す.

我々は3次元コンピュータグラフィックスの技術を用いることで, PC上でポップアップカードの設計を対話的に行えるソフトウェアの研究開発を行ってきた.このソフトウェアを用いることで, PCの画面に表示されるポップアップカードの立体図をマウスとキーボードのカーソルキーで編集し, その展開図をプリンタから出力することができる.この展開図を元にカッターなどを用いて紙工作することで, PCの画面で設計したポップアップカードを実際に手に触れて確認できるようになる.本稿では, このソフトウェアを初等および中等教育における図工や技術・家庭科を対象とした新しい教材として活用することを提案する.中学校の技術・家庭の授業で試行的に利用してもらうために幾つかの機能拡張を行い, そして実際に授業で試行することによって教材としての評価を行った.

Computer Tomography (CT) has been used for inspection purposes in the industry today. It is also used for measuring shapes of mechanical parts. In order to generate shape information from CT data, it is general to use Marching Cubes (MC) method which reconstructs triangular mesh models (MC mesh). Nowadays, because of the high resolution of CT devices, MC meshes become so dense and large to process on ordinary PCs. We propose a method to convert such MC meshes into the form of triangular strips so that they can be compressed and efficiently displayed. The method is based on characteristics particular to the MC meshes, that is, the vertices of the MC meshes are located on the three dimensional grids. In this paper, we describe this method to generate strips from MC meshes and also show experimental results using a realistic example to prove the effectivity of the method.

When creating a physical model to 3D print, the density distribution of an object is important because it determines the mass properties of objects such as center of mass, total mass and moment of inertia. In this paper, we present a density aware shape modelling method to control the mass properties of 3D printed objects. We generate a continuous density distribution that satisfies the given mass properties and generate a 3D printable model that represents this density distribution using a truss structure. The number of nodes and their positions are iteratively optimized so as to minimize error between the target density and the density of the truss structure. With our technique, 3D printed objects that have desired mass properties can be fabricated.

We propose a CT image segmentation method using structural analysis. The aim of our research is to decompose assembled fossil of skeletons and crania into fragments in the area of fossil reconstruction. One challenge specific to this type of segmentation procedure is the separation of fragments where their gaps are not necessarily clear. We previously proposed a method of segmenting CT images using structural analysis. This technique is based on the assumption that the interference area (joint) between components (bones) is structurally weak. We compute strain, which tends to be large in structurally weak areas and segment the image in the region of high strain. With this approach, there is a need to specify boundary conditions for the structural analysis, namely, loading conditions (loading forces and their positions) and locations of fixed boundaries. In our previous work, we proposed a method of optimizing loading forces given loading positions and fixed boundary positions. In this study, we propose a method to find both of those positions to automate the segmentation procedure. Some segmentation results generated by our prototype software demonstrate applicability of the proposed method.

We present a method for computing surface models from MRI images with specific design for vocal tract extraction. As the vocal tract consists of empty space with connection to the outside, it is difficult to segment with common methods based on consideration of intensity differences. In the proposed method, surfaces are extracted via three steps. First, void voxels in an image of a human head are extracted using a morphological closing operator. Then, the significant and unnecessary noise contained in the results is removed via connected-component labeling. Finally, a vocal tract mesh is formed by applying the Marching Cubes algorithm in the extracted region and used to provide masked images. As the method does not require user input, it is efficient for batch processing.

This paper describes developments in sketch input, with particular reference to two sketching applications: building triangulated mesh models for use in computer graphics, and producing boundary representation models of solid objects for use within a CAD system. Both are driven by the same stroke capture mechanism. The mesh modelling system interprets a sketch by reference to a topological template. Hidden parts are generated using the template and assumptions of symmetry and axis-alignment. This tool is more useful for capturing curved shapes which conform to a template topology. The B-rep modelling system makes hypotheses about hidden topology and object geometry based on an estimation of the frontal geometry. The completed topology is chosen by a depth-first search through the tree of hypotheses, and the geometry is finalised using geometric constraint reasoning. This tool is limited to polyhedral shapes without holes, but can cope with a wider range of topologies.

Three-dimensional metamorphosis (or morphing) establishes a smooth transition from a source object to a target object. The primary issue in 3D metamorphosis is to establish surface correspondence between the source and target objects, by which each point on the surface of the source object maps to a point on the surface of the target object. Having established this correspondence, we can generate a smooth transition by interpolating corresponding points from the source to the target positions. We handle 3D geometric metamorphosis between two objects represented as triangular meshes. To improve the quality of 3D morphing between two triangular meshes, we particularly consider the following two issues: 1) metamorphosis of arbitrary meshes; 2) metamorphosis with user control. We can address the first issue using our recently proposed method based on harmonic mapping (T. Kanai et al., 1998). In that earlier work, we developed each of the two meshes (topologically equivalent to a disk and having geometrically complicated shapes), into a 2D unit circle by harmonic mapping. Combining those two embeddings produces surface correspondence between the two meshes. However, this method doesn't consider the second issue: how to let the user control surface correspondence. The article develops an effective method for 3D morphing between two arbitrary meshes of the same topology. We extend our previously proposed method to achieve user control of surface correspondence.