• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.6 no.5
• /
• pp.783-788
• /
• 2002

Recently, research and developmental direction of communication system is concurrent adopting voice data and face image in speaking to provide more higher recognition rate then in the case of only voice data. Therefore, we present a method of lipreading in speech image sequence by using the 3-D facial shape model. The method use a feature information of the face image such as the opening-level of lip, the movement of jaw, and the projection height of lip. At first, we adjust the 3-D face model to speeching face Image sequence. Then, to get a feature information we compute variance quantity from adjusted 3-D shape model of image sequence and use the variance quality of the adjusted 3-D model as recognition parameters. We use the intensity inclination values which obtaining from the variance in 3-D feature points as the separation of recognition units from the sequential image. After then, we use discrete HMM algorithm at recognition process, depending on multiple observation sequence which considers the variance of 3-D feature point fully. As a result of recognition experiment with the 8 Korean vowels and 2 Korean consonants, we have about 80% of recognition rate for the plosives md vowels.

• Geomechanics and Engineering
• /
• v.17 no.4
• /
• pp.317-322
• /
• 2019

This paper presents the results of parametric analyses to compute the axial capacity of a suction pile using 2D and 3D finite element approaches. The study is intended to simplify the process of analyzing suction piles from 3D to 2D model. The research focuses on obtaining the coefficient to be applied into the 2D model in order to obtain results that are as close as possible to the 3D model. Two 2D models were used in the analysis, namely the plane strain and axisymmetric models. The analyses were performed using two actual offshore soil data of the North and West Java Indonesia. The study reveals that the simplification of model through 2D Finite Element is achievable by applying the appropriate coefficient to the stiffness parameters. The results show that the simplified model of the 2D FEA provides more conservative results (with the difference between 2% to 7%) than the 3D FEA.

• Journal of the Society of Naval Architects of Korea
• /
• v.59 no.4
• /
• pp.235-242
• /
• 2022

As ship condition inspection technology has been developed, research on collecting, analyzing, and diagnosing condition information has become active. In ships, related research has been conducted, such as analyzing, detecting, and classifying major hull failures such as cracks and corrosion using 2D and 3D data information. However, for geometric deformation such as indents and bulges, 2D data has limitations in detection, so 3D data is needed to utilize spatial feature information. In this study, we aim to detect hull structural deformation positions. It builds a specimen based on actual hull structure deformation and acquires a point cloud from a model scanned with a 3D scanner. In the obtained point cloud, deformation(outliers) is found with a combination of RANSAC algorithms that find the best matching model in the Octree data structure and dataset.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.6
• /
• pp.591-598
• /
• 2020

The all revision of the Road Name Address Act, which contains the contents to be used by expanding the road name address as a means of indicationg the location, has been resloved by the National Assembly. Addresses will be assigned to large-sized facilities (3D mixed-use complex spaces). Here, the 3D (Three-dimensional) address is assigned an indoor path section in the inner passage, dividing the section at intervals. The 3D address will be built on the address information map. For 3D address, data should be built and managed for a 3D complex space(indoor space). Therefore, in this study, the object of the 3D address is defined based on the address conceptual model defined in the international standard, and the 3D address data model is proposed based on IndoorGML. To this, it is proposed as a method of mapping the Core and Navigation module of IndoorGML so that the entity of the 3D address can be expressed in IndoorGML. This study has a limitation in designing a 3D address data model only, but it is meaningful that it suggested a standard for constructing 3D address data in the future.

• Current Optics and Photonics
• /
• v.2 no.6
• /
• pp.554-562
• /
• 2018

This study presents a color 3D scanner based on the laser structured-light imaging method that can simultaneously acquire 3D shape data and color of a target object using a single camera. The 3D data acquisition of the scanner is based on the structured-light imaging method, and the color data is obtained from a natural color image. Because both the laser image and the color image are acquired by the same camera, it is efficient to obtain the 3D data and the color data of a pixel by avoiding the complicated correspondence algorithm. In addition to the 3D data, the color data is helpful for enhancing the realism of an object model. The proposed scanner consists of two line lasers, a color camera, and a rotation table. The line lasers are deployed at either side of the camera to eliminate shadow areas of a target object. This study addresses the calibration methods for the parameters of the camera, the plane equations covered by the line lasers, and the center of the rotation table. Experimental results demonstrate the performance in terms of accurate color and 3D data acquisition in this study.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 1998.11a
• /
• pp.181-185
• /
• 1998

Recently, research and developmental direction of communication system is concurrent adopting voice data and face image in speaking to provide more higher recognition rate then in the case of only voice data. Therefore, we present a method of lipreading in speech image sequence by using the 3-D facial shape model. The method use a feature information of the face image such as the opening-level of lip, the movement of jaw, and the projection height of lip. At first, we adjust the 3-D face model to speeching face image sequence. Then, to get a feature information we compute variance quantity from adjusted 3-D shape model of image sequence and use the variance quality of the adjusted 3-D model as recognition parameters. We use the intensity inclination values which obtaining from the variance in 3-D feature points as the separation of recognition units from the sequential image. After then, we use discrete HMM algorithm at recognition process, depending on multiple observation sequence which considers the variance of 3-D feature point fully. As a result of recognition experiment with the 8 Korean vowels and 2 Korean consonants, we have about 80% of recognition rate for the plosives and vowels.

• Spatial Information Research
• /
• v.21 no.2
• /
• pp.11-21
• /
• 2013

W ith the recent increasing build and application for 3D spatial information, the importance of management and application for spatial information based on indoor space has been increased. Especially, Due to the increasing of the scale and complexity of the building according to the development of construction technologies several studies have been conducted to provide the services based on indoor space such as indoor navigation for disaster. Therefore, to efficient manage and service for information of complicated indoor space, it is necessary to extend and develop 3D spatial model and services that have been developed for outdoor space. In this paper, Indoor Spatial Data Model(ISDM) is developed to support building spatial information for complicated indoor space and location based services through topological information. ISDM contains a feature model which is a CityGML Application Domain Extension(ADE) model and a topology model that refers the IndoorGML.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1250-1256
• /
• 2019

Although one can easily generate real-world 3D mesh data using a 3D printer or a depth camera, the generated data inevitably includes unnecessary noise. Therefore, mesh denoising is essential to obtain intact 3D mesh data. However, conventional mathematical denoising methods require preprocessing and often eliminate some important features of the 3D mesh. To address this problem, this paper proposes a deep learning based 3D mesh denoising method. Specifically, we propose a convolution-based autoencoder model consisting of an encoder and a decoder. The convolution operation applied to the mesh data performs denoising considering the relationship between each vertex constituting the mesh data and the surrounding vertices. When the convolution is completed, a sampling operation is performed to improve the learning speed. Experimental results show that the proposed autoencoder model produces faster and higher quality denoised data than the conventional methods.

• The KIPS Transactions:PartD
• /
• v.12D no.3 s.99
• /
• pp.375-384
• /
• 2005

Most of data objects in the real world have spatial and temporal information simultaneously. One of the objectives of spatiotemporal databases is to provide a very expressive framework for spatiotemporal database applications. The object oriented spatiotemporal model can manage spatiotemporal data efficiently because the model is able to store the complete history of any objects. Therefore, in order to handle the spatiotemporal data in object oriented fashion, we proposed a low dimensional spatiotemporal model in the object oriented fashion that extended dimensionally two dimensional spatial model of OpenGIS to fit into the real world information and to include temporal information. In order to proof the proposed model, we design a data provider component which consists of the data source, the session, the command, and the rowset component, and implement it. The proposed four dimensional spatiotemporal data model and provider in the object oriented fashion can be used to store, retrieve and manage the object oriented spatiotemporal data.

• Proceedings of the Korean Society of Precision Engineering Conference
• /
• 2002.05a
• /
• pp.536-539
• /
• 2002

The reverse engineering (RE) technology can quickly generate 3D point cloud data of an object by capturing the surface of a model using a 3D scanner. In the rapid prototyping (RP) technology, prototypes are rapidly produced from 3D CAD models in a layer-by-layer additive basis. In this paper, a physical human head shape is duplicated using a new RP process, the Transfer-type Variable Lamination Manufacturing process using expandable polystyrene foam sheet (VLM-ST), after the point cloud data of a human head shape measured from 3D SNX scanner are converted to STL file. From the duplicated human head shape, it has been shown that the VLM-ST process in connection with the 3D scanner is a fast and efficient process in that shapes with free surface, such as the human head shape, can be duplicated with ease. Considering the measurement time and the shape duplication time, the use of 3D SNX scanner and the VLM-ST process is expected to reduce the lead-time fur the development of new products in comparison with the other existing RE-RP connected manufacturing systems.