• Journal of IKEEE
• /
• v.26 no.2
• /
• pp.160-168
• /
• 2022

In this paper, we propose a 3D mesh reconstruction method from a single image using deep learning and a sphere shape transformation method. The proposed method has the following originality that is different from the existing method. First, the position of the vertex of the sphere is modified to be very similar to the 3D point cloud of an object through a deep learning network, unlike the existing method of building edges or faces by connecting nearby points. Because 3D point cloud is used, less memory is required and faster operation is possible because only addition operation is performed between offset value at the vertices of the sphere. Second, the 3D mesh is reconstructed by covering the surface information of the sphere on the modified vertices. Even when the distance between the points of the 3D point cloud created by correcting the position of the vertices of the sphere is not constant, it already has the face information of the sphere called face information of the sphere, which indicates whether the points are connected or not, thereby preventing simplification or loss of expression. can do. In order to evaluate the objective reliability of the proposed method, the experiment was conducted in the same way as in the comparative papers using the ShapeNet dataset, which is an open standard dataset. As a result, the IoU value of the method proposed in this paper was 0.581, and the chamfer distance value was It was calculated as 0.212. The higher the IoU value and the lower the chamfer distance value, the better the results. Therefore, the efficiency of the 3D mesh reconstruction was demonstrated compared to the methods published in other papers.

• Journal of Broadcast Engineering
• /
• v.27 no.4
• /
• pp.548-560
• /
• 2022

Video based Point Cloud Compression (V-PCC) is one of the compression methods for compressing point clouds, and shows high efficiency in dynamic point cloud compression with movement due to the feature of compressing point cloud data using an existing video codec. Accordingly, V-PCC is drawing attention as a core technology for immersive content services such as AR/VR. In order to effectively service these V-PCC contents through a media streaming platform, it is necessary to encapsulate them in the existing media file format, ISO based Media File Format (ISOBMFF). However, in order to service through an adaptive streaming platform such as Dynamic Adaptive Streaming over HTTP (DASH), it is necessary to encode V-PCC contents of various qualities and store them in the server. Due to the size of the 2D media, it causes a great burden on the encoder and the server compared to the existing 2D media. As a method to solve such a problem, it may be considered to configure a streaming platform based on content obtained through V-PCC content encoding based on SHVC. Therefore, this paper encapsulates the SHVC-based V-PCC bitstream into ISOBMFF suitable for DASH service and proposes a configuration method to service it. In addition, in this paper, we propose ISOBMFF encapsulation and DASH configuration method to effectively service SHVC-based V-PCC contents, and confirm them through verification experiments.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.9
• /
• pp.1330-1339
• /
• 2022

3D object detection generally aims to detect relatively large data such as automobiles, buses, persons, furniture, etc, so it is vulnerable to small object detection. In addition, in an environment with limited resources such as embedded devices, it is difficult to apply the model because of the huge amount of computation. In this paper, the accuracy of small object detection was improved by focusing on local features using only one layer, and the inference speed was improved through the proposed knowledge distillation method from large pre-trained network to small network and adaptive quantization method according to the parameter size. The proposed model was evaluated using SUN RGB-D Val and self-made apple tree data set. Finally, it achieved the accuracy performance of 62.04% at mAP@0.25 and 47.1% at mAP@0.5, and the inference speed was 120.5 scenes per sec, showing a fast real-time processing speed.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.154-156
• /
• 2021

최근 3차원 영상이 다양한 분야에서 활용되고 있으며, 이에 따라 3차원 영상에 대한 압축과 전송 방안에 대한 연구가 활발히 진행되고 있다. 국제 표준화 기구인 ISO/IEC 산하 Moving Picture Expert Group(MPEG)에서는 기존의 2차원 비디오 코덱을 이용하여 고밀도 포인트 클라우드 압축하는 방안인 V-PCC와 3DoF+ 영상을 압축하기 위한 방안인 MPEG Immersive Video(MIV)를 표준화 중에 있다. V-PCC와 MIV는 압축 방법의 유사성으로 인해 동일한 Volumetric Visual Video-based Coding(V3C) 형식으로 저장된다. 압축된 V3C 데이터를 효과적으로 저장하여 이용하기 위해서는 ISO based Media File Format(ISOBMFF) 캡슐화 과정이 필수적이다. 본 논문에서는 MPEG의 Carriage of V3C data 표준에 따라 V3C 데이터를 ISOBMFF로 캡슐화 실험을 진행하였으며, 실험에 대한 검증을 위하여 생성된 ISOBMFF 데이터를 V3C 데이터로 복원한 뒤, 디코딩 하여 확인하였다.

• Annual Conference of KIPS
• /
• 2023.05a
• /
• pp.298-300
• /
• 2023

자율주행 및 robot navigation 시스템에서 물체 인식 성능향상을 위해 대부분 MSF(Multi-Sensor Fusion) 기반 설계를 한다. 따라서 각 센서로부터 들어온 정보를 정합하는 것은 정확한 MSF 알고리즘을 위한 필요조건이다. 다양한 선행 연구에서 2D 데이터에 대한 공격을 진행했다. 자율주행에서는 3D 데이터를 다루어야 하므로 선행 연구에서 하지 않았던 3D 데이터 공격을 진행했다. 본 연구에서는 스케일링 공격 기반 카메라-라이다 센서 간 정합 모델의 정확도를 저하시키는 공격 방법을 제안한다. 제안 방법은 입력 라이다의 포인트 클라우드에 스케일링 공격을 적용하여 다운스케일링 단계에서 공격하고자 한다. 실험 결과, 입력 데이터에 공격하였을 때 공격 전보다 평균제곱 이동오류는 56% 이상, 평균 사원수 각도 오류는 98% 이상 증가했음을 보였다. 다운스케일링 크기 별, 알고리즘별 공격을 적용했을 때, 10×20 크기로 다운스케일링 하고 lanczos4 알고리즘을 적용했을 때 가장 효과적으로 공격할 수 있음을 확인했다.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.6
• /
• pp.363-369
• /
• 2021

High-definition road maps are used as the basic infrastructure for autonomous vehicles, so the latest road information must be quickly reflected. However, the current drawing and structural editing process of high-definition road maps are manually performed. In addition, it takes the longest time to generate road lanes, which are the main construction targets. In this study, the point cloud of the road lane markings, in which color types(white, blue, and yellow) were predicted through the PointNet model pre-trained in previous studies, were used as input data. Based on the point cloud, this study proposed a methodology for automatically drawing and structural editing of the layer of road lane markings. To verify the usability of the 3D vector data constructed through the proposed methodology, the accuracy was analyzed according to the quality inspection criteria of high-definition road maps. In the positional accuracy test of the vector data, the RMSE (Root Mean Square Error) for horizontal and vertical errors were within 0.1m to verify suitability. In the structural editing accuracy test of the vector data, the structural editing accuracy of the road lane markings type and kind were 88.235%, respectively, and the usability was verified. Therefore, it was found that the methodology proposed in this study can efficiently construct vector data of road lanes for high-definition road maps.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.11
• /
• pp.224-229
• /
• 2018

Korea is experiencing a rapid increase in electricity consumption due to rapid economic development, and many power transmission towers are installed to provide smooth power supply. The high-voltage transmission line is mainly made of aluminum stranded wire, and the wire is loosely guided so that some deflection is maintained. The degree of deflection has a great influence on the quality of the construction and the life of the cable. As the time passes, the shrinkage and expansion occur repeatedly due to the weight of the cable and the surrounding environment. Therefore, periodic monitoring is essential for the management of the power transmission line. In this study, the power transmission lines were monitored using 3D laser scanning technology. The data of the power transmission line of the study area was acquired and the point cloud type 3D geospatial information of the transmission line was extracted through data processing. The length of the transmission line and deflection amount were calculated using the 3D geospatial information of the transmission line, and the distance from the surrounding obstacles could be calculated effectively. The result of study shows the utilization of 3D laser scanning technology for transmission line management. Future research will contribute to the efficiency of transmission line management if a transmission line monitoring system using 3D laser scanning technology is developed.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2022.07a
• /
• pp.159-160
• /
• 2022

본 논문에서는 최근 주요시설의 경계에 주로 사용되기 시작한 라이다 센서를 여러대 사용할때 보다 효율적으로 사용하기 위해서 통합된 3차원 좌표계에서 시각화하는 방법에 대해 설명한다. 주로 카메라 기반 CCTV의 경우 정확성은 높지만 시야각(Field of View)이 좁기 때문에 레이더(RADAR)센서와 같은 센서와 함께 혼용되는 경우가 많다. 레이더 센서의 데이터는 넓은 범위에 대한 감지를 할 수 있지만 노이즈가 많고 물체의 형상을 정확하게 측정하기 힘들다. 라이다(LiDAR) 센서는 레이져를 이용하여 멀고 넓은 범위를 정교하게 측정할 수 있다. 이러한 라이다 센서는 정교한 만큼 처리해야할 데이터의 양이 많으며, 다수의 센서를 이용하더라도 하나의 화면에서 처리하기 힘들다는 단점이 있다. 제안하는 논문은 여러개의 라이다 센서에서 측정한 데이터를 실시간에 하나의 좌표계로 통일하여 하나의 영상을 보일 수 있도록 통합 뷰잉 환경을 제공한다.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.3
• /
• pp.133-139
• /
• 2021

Currently, the process of constructing a high-definition road map has a high proportion of manual labor, so there are limitations in construction time and cost. Research to automate map production with high-definition road maps using artificial intelligence is being actively conducted, but since the construction of training data for the map construction is also done manually, there is a need to automatically build training data. Therefore, in this study, after converting to images using point clouds acquired by a mobile mapping system, the road marking areas were extracted through image reclassification and overlap analysis using thresholds. Then, a methodology was proposed to automatically construct training data for deep learning data for the high-definition road map through the classification of the polygon types in the extracted regions. As a result of training 2,764 lane data constructed through the proposed methodology on a deep learning-based PointNet model, the training accuracy was 99.977%, and as a result of predicting the lanes of three color types using the trained model, the accuracy was 99.566%. Therefore, it was found that the methodology proposed in this study can efficiently produce training data for high-definition road maps, and it is believed that the map production process of road markings can also be automated.

• Proceedings of the KIEE Conference
• /
• 2015.07a
• /
• pp.1373-1374
• /
• 2015

본 논문은 복잡한 비선형 모델링 방법인 다항식 기반 RBF 뉴럴 네트워크(Radial Basis Function Neural Network)와 벡터공간에서 임의의 비선형 경계를 찾아 두 개의 집합을 분류하는 방법으로 주어진 조건하에서 수학적으로 최적의 해를 찾는 SVM(Support Vector Machine)를 사용하여 3차원 얼굴인식 모델을 설계하고 두 모델의 3차원 얼굴 인식률을 비교한다. 3D스캐너를 통해 3차원 얼굴형상을 획득하고 획득한 영상을 전처리 과정에서 포인트 클라우드 정합과 포즈보상을 수행한다. 포즈보상 통해 정면으로 재배치한 영상을 Multiple Point Signature기법을 이용하여 얼굴의 깊이 데이터를 추출한다. 추출된 깊이 데이터를 RBFNN과 SVM의 입력패턴과 출력으로 선정하여 모델을 설계한다. 각 모델의 효율적인 학습을 위해 PCA 알고리즘을 이용하여 고차원의 패턴을 축소하여 모델을 설계하고 인식 성능을 비교 및 확인한다.