• Journal of the Society of Naval Architects of Korea
• /
• v.59 no.5
• /
• pp.330-337
• /
• 2022

Research on object detection algorithms using 2D data has already progressed to the level of commercialization and is being applied to various manufacturing industries. Object detection technology using 2D data has an effective advantage, there are technical limitations to accurate data generation and analysis. Since 2D data is two-axis data without a sense of depth, ambiguity arises when approached from a practical point of view. Advanced countries such as the United States are leading 3D data collection and research using 3D laser scanners. Existing processing and detection algorithms such as ICP and RANSAC show high accuracy, but are used as a processing speed problem in the processing of large-scale point cloud data. In this study, PointNet a representative technique for detecting objects using widely used 3D point cloud data is analyzed and described. And RandLA-Net, which overcomes the limitations of PointNet's performance and object prediction accuracy, is described a review of detection technology using point cloud data was conducted.

• Journal of Broadcast Engineering
• /
• v.26 no.6
• /
• pp.680-691
• /
• 2021

Due to the recent developments of attaining 3D contents by using devices such as 3D scanners, the diversity of the contents being used in AR(Augmented Reality)/VR(Virutal Reality) fields is significantly increasing. There are several ways to represent 3D data, and using point clouds is one of them. A point cloud is a cluster of points, having the advantage of being able to attain actual 3D data with high precision. However, in order to express 3D contents, much more data is required compared to that of 2D images. The size of data needed to represent dynamic 3D point cloud objects that consists of multiple frames is especially big, and that is why an efficient compression technology for this kind of data must be developed. In this paper, a motion estimation and compensation method for dynamic point cloud objects using 3D DCT is proposed. This will lead to switching the 3D video frames into I frames and P frames, which ensures higher compression ratio. Then, we confirm the compression efficiency of the proposed technology by comparing it with the anchor technology, an Intra-frame based compression method, and 2D-DCT based V-PCC.

• Journal of Broadcast Engineering
• /
• v.26 no.6
• /
• pp.714-724
• /
• 2021

The point cloud content is immersive content recorded by acquiring points and colors corresponding to the real environment and objects having three-dimensional location information. When a point cloud content consisting of three-dimensional points having position and color information is enlarged and rendered, the gap between the points widens and an empty hole occurs. In this paper, we propose a method for improving the quality of point cloud contents through inverse transformation-based interpolation using depth information for holes by finding holes that occur due to the gap between points when expanding the point cloud. The points on the back are rendered between the holes created by the gap between the points, acting as a hindrance to applying the interpolation method. To solve this, remove the points corresponding to the back side of the point cloud. Next, a depth map at the point in time when an empty hole is generated is extracted. Finally, inverse transform is performed to extract pixels from the original data. As a result of rendering content by the proposed method, the rendering quality improved by 1.2 dB in terms of average PSNR compared to the conventional method of increasing the size to fill the blank area.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.04a
• /
• pp.896-897
• /
• 2015

본 연구는 3D 이미지 스캔 데이터 기반으로, SWEEPING 형상을 효과적으로 역설계하는 기술에 관한 것이다. 사용자가 미리 정의한 형상 단면 모델 데이터베이스를 이용해, 3차원 SWEEPING 형상을 자동으로 역설계하는 알고리즘을 제안한다. 이를 위해, 3D 이미지 스캔 데이터인 포인트 클라우드에서 자동으로 추출한 단면 포인트들을 처리해, 파라메터 정보를 추출하고, 미리 정의된 형상 단면들과 상호간 유사도를 비교한 후, 가장 유사한 형상 단면을 획득한다. 이러한 기술은 SWEEPING 형상 모델의 역설계 과정을 자동화하는 데 도움을 줄 것이다.

• Journal of IKEEE
• /
• v.26 no.3
• /
• pp.408-415
• /
• 2022

In this paper, we propose a 3D point cloud reconstruction technique from 2D images using efficient feature map extraction network. The originality of the method proposed in this paper is as follows. First, we use a new feature map extraction network that is about 27% efficient than existing techniques in terms of memory. The proposed network does not reduce the size to the middle of the deep learning network, so important information required for 3D point cloud reconstruction is not lost. We solved the memory increase problem caused by the non-reduced image size by reducing the number of channels and by efficiently configuring the deep learning network to be shallow. Second, by preserving the high-resolution features of the 2D image, the accuracy can be further improved than that of the conventional technique. The feature map extracted from the non-reduced image contains more detailed information than the existing method, which can further improve the reconstruction accuracy of the 3D point cloud. Third, we use a divergence loss that does not require shooting information. The fact that not only the 2D image but also the shooting angle is required for learning, the dataset must contain detailed information and it is a disadvantage that makes it difficult to construct the dataset. In this paper, the accuracy of the reconstruction of the 3D point cloud can be increased by increasing the diversity of information through randomness without additional shooting information. In order to objectively evaluate the performance of the proposed method, using the ShapeNet dataset and using the same method as in the comparative papers, the CD value of the method proposed in this paper is 5.87, the EMD value is 5.81, and the FLOPs value is 2.9G. It was calculated. On the other hand, the lower the CD and EMD values, the better the accuracy of the reconstructed 3D point cloud approaches the original. In addition, the lower the number of FLOPs, the less memory is required for the deep learning network. Therefore, the CD, EMD, and FLOPs performance evaluation results of the proposed method showed about 27% improvement in memory and 6.3% in terms of accuracy compared to the methods in other papers, demonstrating objective performance.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2019.06a
• /
• pp.234-235
• /
• 2019

본 논문은 실사 객체를 360도 전방위에서 관찰이 가능한 3D 그래픽 모델로 변환하는 시스템에서 뼈대를 추출하는 방법을 제시한다. 각 카메라로부터 촬영된 텍스쳐 영상을 이용하여 뼈대를 추출하고, 깊이 정보로부터 얻어진 포인트 클라우드 정보를 이용하여 뼈대 정보를 정합, 보정하는 과정을 수행한다. 카메라로부터 촬영된 텍스쳐 영상에 대해 딥러닝 기술 등을 이용하여 뼈대를 획득한다. 텍스쳐 영상으로부터 획득된 뼈대 정보는 동일 위치에서 획득된 외부 파라미터를 이용하여 월드좌표계로 변환하여 공간상에 위치시킨다. 이러한 과정을 모든 카메라로부터 획득된 뼈대 정보에 동일하게 적용함으로써 모든 뼈대 정보를 공간상에 표현하여 최종적인 뼈대 정보를 추출하는 방법을 제시한다.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.10a
• /
• pp.10-11
• /
• 2015

최근 자율 주행 분야의 연구에서 LIDAR를 활용한 분류 기법들이 연구되고 있다. 그러나 2D 영상 처리와 비교하여, 대량의 3D 포인트를 사용하는 분류 알고리즘의 성능을 평가하기 위한 지상 검증자료를 쉽게 획득하기 어렵다. 본 연구는 LIDAR를 가상 공간에서 시뮬레이션 함으로써 감지한 물체의 정보를 기록함으로써 3D 포인트 클라우드를 사용하는 다양한 분류 알고리즘의 검증을 위한 지상검증자료를 생성하는 기법을 설명한다. 본 기법은 실제 LIDAR를 사용하는 것보다 적은 비용으로 다양한 환경에서의 분류 알고리즘 성능 검증을 가능하게 한다.

• Journal of KIISE
• /
• v.43 no.5
• /
• pp.562-568
• /
• 2016

Point cloud data can be expressed in a specific coordinate system of a data set with a large number of points, to represent any form that generally has different characteristics in the three-dimensional coordinate space. This paper is aimed at finding a cylindrical pipe in the point cloud of the three-dimensional coordinate system using RANSAC, which is faster than the conventional Hough Transform method. In this study, the proposed cylindrical pipe is estimated by combining the results of parameters based on two mathematical models. The two kinds of mathematical models include a sphere and line, searching the sphere center point and radius in the cylinder, and detecting the cylinder with straightening of center. This method can match cylindrical pipe with relative accuracy; furthermore, the process is rapid except for normal estimation and segmentation. Quick cylinders matching could benefit from laser scanning and reverse engineering construction sectors that require pipe real-time estimates.

• Journal of Digital Convergence
• /
• v.18 no.11
• /
• pp.427-432
• /
• 2020

In order to establish an effective forest management plan, it is necessary to investigate tree management information such as tree height and DBH(Diameter at breast height). However, research on convergence and application of data acquisition technology to improve the efficiency of existing forest survey methods is insufficient. Therefore, in this study, tree management information was constructed and analyzed using point cloud data acquired through a 3D scanner. Data on the study site was acquired using fixed and mobile 3D scanners, and the efficiency of the mobile 3D scanner was presented through comparison of working hours. In addition, tree management information for object management was constructed by classifying vegetation by object using point cloud data, and by constructing information on chest height diameter and height. As a result of the accuracy evaluation compared with the conventional measurement method, the difference in tree height was 0.02-0.09m and DBH was 0.01-0.04m. If information on the location of vegetation and crowns of each object is constructed through additional research in the future, the efficiency of the work related to forest management information construction can be greatly increased.

• Journal of Broadcast Engineering
• /
• v.24 no.5
• /
• pp.849-858
• /
• 2019

Recent developments in computer graphics technology and image processing technology have increased interest in point cloud technology for inputting real space and object information as three-dimensional data. In particular, point cloud technology can accurately provide spatial information, and has attracted a great deal of interest in the field of autonomous vehicles and AR (Augmented Reality)/VR (Virtual Reality). However, in order to provide users with 3D point cloud contents that require more data than conventional 2D images, various technology developments are required. In order to solve these problems, an international standardization organization, MPEG(Moving Picture Experts Group), is in the process of discussing efficient compression and transmission schemes. In this paper, we provide a region division transfer method of 3D point cloud content through extension of existing MPEG-DASH (Dynamic Adaptive Streaming over HTTP)-SRD (Spatial Relationship Description) technology, quality parameters are further defined in the signaling message so that the quality parameters can be selectively determined according to the user's request. We also design a verification platform for ROUTE (Real Time Object Delivery Over Unidirectional Transport)/DASH based heterogeneous network environment and use the results to validate the proposed technology.