• International Journal of Internet, Broadcasting and Communication
• /
• v.13 no.2
• /
• pp.208-217
• /
• 2021

3D-CNN is one of the deep learning techniques for learning time series data. However, these three-dimensional learning can generate many parameters, requiring high performance or having a significant impact on learning speed. We will use these 3D-CNNs to learn hand gesture and find the parameters that showed the highest accuracy, and then analyze how the accuracy of 3D-CNN varies through input data changes without any structural changes in 3D-CNN. First, choose the interval of the input data. This adjusts the ratio of the stop interval to the gesture interval. Secondly, the corresponding interframe mean value is obtained by measuring and normalizing the similarity of images through interclass 2D cross correlation analysis. This experiment demonstrates that changes in input data affect learning accuracy without structural changes in 3D-CNN. In this paper, we proposed two methods for changing input data. Experimental results show that input data can affect the accuracy of the model.

• Journal of the Korean Society of Manufacturing Process Engineers
• /
• v.19 no.9
• /
• pp.59-70
• /
• 2020

Three-dimensional (3D) design data is used for various purposes throughout the life cycle of a plant construction project. Plant 3D CAD systems support 3D modeling based on specs-catalogs, which contain data that are used for different purposes such as design, procurement, production, and handover. Therefore, it is important to share the spec-catalog data in the 3D design model with other application systems. Sharing this data thus requires a system that extracts spec-catalog data from plant 3D CAD systems and converts them into neutral model data. In this paper, we analyze equipment spec-catalog data of plant 3D CAD systems and, based on these analyses, define the data structure for neutral spec-catalog data. We subsequently propose a procedure that translates native spec-catalog data to neutral model data and develop a prototype system that performs this operation. The proposed method is then experimentally validated for the test spec-catalog data.

• Journal of Institute of Control, Robotics and Systems
• /
• v.11 no.9
• /
• pp.768-773
• /
• 2005

A new approach for human foot modelling and last design based on the cross sectional method is presented in this paper. The proposed last design method utilizes the dynamic trimmed parametric patches for the foot 3D data and last 3D data. The cross section a surface of 3D foot for the 3D last, design modeling of free form geometric last shapes. The proposed last design scheme wraps the 3D last data surrounding the measured 3D foot data with the effect of deforming the last design rule The last design rule of the FFD is constructed on the FFD lattice based on foot-last shape analysis. In addition, the control points of FFD lattice are constructed with cross sectional data interpolation methods from the a finite set of 3D foot data. The deformed 3D last result obtained from the proposed FFD is saved as a 3D dxf foot data. The experimental results demonstrate that the last designed with the proposed scheme has good performance.

• Journal of Institute of Control, Robotics and Systems
• /
• v.9 no.9
• /
• pp.700-706
• /
• 2003

This paper presents a 3D last design system that provides the 3-dimensional last data based on the FFD(Free Form Deformation) method. The proposed system utilizes the control points for deformation factor to convert from the 3D point cloud foot data to the 3D point cloud last data. The deformation factor of the FFD is obtained from the conventional last design technique, and constructed on the FFD lattice based on the bottom view and lateral view of the measured 3D point cloud foot data. In addition, the control points of FFD lattice is decided on the anatomical points of foot. The deformed 3D last obtained from the proposed FFD is saved as a 3D dxf foot data. The experimental results demonstrate that the proposed system have the descent 3D last data based on the openGL window.

• Journal of Internet Computing and Services
• /
• v.23 no.6
• /
• pp.71-78
• /
• 2022

This paper proposes an algorithm for 3D object detection by processing point cloud data of 3D LiDAR. Unlike 2D LiDAR, 3D LiDAR-based data was too vast and difficult to process in three dimensions. This paper introduces various studies based on 3D LiDAR and describes 3D LiDAR data processing. In this study, we propose a method of processing data of 3D LiDAR using clustering techniques for object detection and design an algorithm that fuses with cameras for clear and accurate 3D object detection. In addition, we study models for clustering 3D LiDAR-based data and study hyperparameter values according to models. When clustering 3D LiDAR-based data, the DBSCAN algorithm showed the most accurate results, and the hyperparameter values of DBSCAN were compared and analyzed. This study will be helpful for object detection research using 3D LiDAR in the future.

• Journal of the Computational Structural Engineering Institute of Korea
• /
• v.22 no.5
• /
• pp.475-485
• /
• 2009

In this paper a two-step algorithm is proposed to efficiently generate rebar elements from 3D CAD data in the context of CAD/CAE data transfer. The first step is an algorithm to identify various type rebar objects and their attributes by analyzing 3D CAD data in STEP format, which is one of the international data standards. The second algorithmic step is a procedure to generate one-dimensional rebar elements from the object data made through the first step for finite element analysis or other CAE tasks. Successful rebar element data generation from real 3D CAD data for a reinforced concrete structure shows the efficacy of the proposed algorithm.

• Korean Journal of Human Ecology
• /
• v.19 no.3
• /
• pp.555-563
• /
• 2010

With recent advances in 3D scanning technology, three-dimensional (3D) patternmaking is becoming a powerful way to develop garments pattern. This technology is now applicable to the made to measure (MTM) system of both ordinary and tightly fitting garments. Although the pattern of fitted clothing has been developed using 3D human data, it is still interesting to develop cycling pants by considering while-cycling body posture and fabric elasticity. This study adopted the Garland's triangle simplification method in order to simplify data without distorting the original 3D scan. Next, the Runge-Kutta method (2C-AN program) was used to develop a 2D pattern from the triangular pixels in the 3D scanned data. The 3D scanned data of four male, university students aged from 21 to 25, was obtained using Whole body scanner (Model WB4, Cyberware, Inc., USA). Results showed the average error of measurement was $4.58cm^2$ (0.19%) for area and 0~0.61cm for the length between the 3D body scanned data and the 2D developed pattern data. This is an acceptable range of error for garment manufacture. Additionally, the 2D pattern developed, based on the 3D body scanned data, did not need ease for comfort or ease of movement when cycling. This study thus provides insights into how garment patterns may be developed for ergonomic comfort in certain special environments.

• Korean Journal of Human Ecology
• /
• v.21 no.5
• /
• pp.975-988
• /
• 2012

Although the pattern development for tight-fitting clothing has been carried out using 3D data on humans, the pattern development using 3D scan data obtained for various postures still remains an interesting subject. In this study, we have developed the 2D pattern using the 3D human body reflecting standing and cycling postures. The 3D scan data of a subject was obtained using Cyberware. 2C-AN program(Triangle simplification and the Runge-Kutta method) was used in the system to reduce the 3D scan data points and to make segmented triangular patches in a plane from 3D data. As results, surface distance and area of each body part of standing and cycling postures were also provided for the future application of the functional clothing construction. The area of center piece on the front (c.front) decreased by $106.45cm^2$(-13.08%) and that of lateral piece(s.back) on the back increased by $144.96cm^2$(18.69%) in the patterns of cycling posture. The girth of neck and waist for the cycling posture increased by 0.88cm (3.92%) and 1.56cm(4.40%) respectively, and the that of thigh decreased by 1.01cm(-2.24%). The differences between the area in the 2D pattern obtained from the 3D scan data and that in the 3D scan surface data for standing and cycling postures were very small($-10.34cm^2$(-0.32%) and $-44.33cm^2$(-1.32%)).

• Journal of Korean Society for Geospatial Information Science
• /
• v.18 no.2
• /
• pp.113-118
• /
• 2010

Recently, 3D-reconstruction for geographic information and study of geospatial information is progressing in various fields through national policy such as R&D business and pilot project. LiDAR system has a advantage of acquisition the 3D information data easily and densely so that is used in many different fields. Considering to characterist of the point data formed with 3D, it need a high specification CPU because it requires a number of processing operation for 2D form expressed by monitor. In contrast, 2D grid structure, like DEM, has a advantage on costs because of simple structure and processing speed. Therefore, purpose of this study is to solve the problem of requirement of more storage space, when LiDAR data stored in forms of 3D is used for 3D-geographic and 3D-buliding representation. Additionally, This study reconstitutes 2D-gird data to supply the representation data of 3D-geographic and presents the storage method which is available for detailed representation applying tree-structure and reduces the storage space.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.16 no.5
• /
• pp.215-223
• /
• 2021

3D data can be categorized into two parts : Euclidean data and non-Euclidean data. In general, 3D data exists in the form of non-Euclidean data. Due to irregularities in non-Euclidean data such as mesh and point cloud, early 3D deep learning studies transformed these data into regular forms of Euclidean data to utilize them. This approach, however, cannot use memory efficiently and causes loses of essential information on objects. Thus, various approaches that can directly apply deep learning architecture to non-Euclidean 3D data have emerged. In this survey, we introduce various deep learning methods for mesh and point cloud data. After analyzing the operating principles of these methods designed for irregular data, we compare the performance of existing methods for shape classification and segmentation tasks.