• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.38 no.4
• /
• pp.345-352
• /
• 2020

To preserve and restore tidal flats and prevent safety accidents, it is necessary to construct tidal flat topographic information including the exact location and shape of tidal creeks. In the tidal flats where the field surveying is difficult to apply, airborne LiDAR surveying can provide accurate terrain data for a wide area. On the other hand, we can economically obtain relatively high-resolution data from UAV (Unmanned Aerial Vehicle) surveying. In this study, we proposed the methodology to generate high-resolution topographic information of tidal flats effectively by integrating airborne LiDAR and UAV point clouds. For the purpose, automatic ICP (Iterative Closest Points) registration between two different datasets was conducted and tidal creeks were extracted by applying CSF (Cloth Simulation Filtering) algorithm. Then, we integrated high-density UAV data for tidal creeks and airborne LiDAR data for flat grounds. DEM (Digital Elevation Model) and tidal flat area and depth were generated from the integrated data to construct high-resolution topographic information for large-scale tidal flat map creation. As a result, UAV data was registered without GCP (Ground Control Point), and integrated data including detailed topographic information of tidal creeks with a relatively small data size was generated.

• Journal of the Korea Institute of Building Construction
• /
• v.19 no.3
• /
• pp.239-246
• /
• 2019

According to the survey on the status of aged buildings in Korea, A number of concrete buildings deterioration such as houses and apartment buildings has been increased rapidly. To solve this problem, the research related to the facility management, that is one of the importance factor, for monitoring buildings has been increased. The research is divided into Survey-based and Technique-based. However, the problem is that Survey-based research is required a lot of time, money and manpower for management. Also, safety cannot be guaranteed in the case of high-rise buildings. Technique-based research has limitations to applying to the current facility maintenance system, as detailed information of deteriorated facilities is difficult to grasp and errors in accuracy are feared. Therefore, this paper contribute to improve the environment of facility management by 4D maps using UAV, camera and Pix4D mapper program to make 3D model. In addition, it is expected to suggest that residents will be offered easy verification to their buildings deterioration.

• Journal of the Korea Society of Computer and Information
• /
• v.28 no.8
• /
• pp.95-101
• /
• 2023

Colmap is one of the innovative artificial intelligence technologies, highly effective as a tool in 3D reconstruction tasks. Moreover, it excels at constructing intricate 3D models by utilizing images and corresponding metadata. Colmap generates 3D models by merging 2D images, camera position data, depth information, and so on. Through this, it achieves detailed and precise 3D reconstructions, inclusive of objects from the real world. Additionally, Colmap provides rapid processing by leveraging GPUs, allowing for efficient operation even within large data sets. In this paper, we have presented a method of collecting 2D images of traditional Korean towers and reconstructing them into 3D models using Colmap. This study applied this technology in the restoration process of traditional stone towers in South Korea. As a result, we confirmed the potential applicability of Colmap in the field of cultural heritage restoration.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.20 no.3
• /
• pp.520-525
• /
• 2019

Mines are an important infrastructure for securing resources, but safety problems can arise in the course of operation. Recently, the mining process is very complicated due to the large scale and mechanization. Therefore, it is necessary to construct accurate geospatial information on mine for systematic and safe mine operation. The geospatial information construction using the existing total station has a disadvantage that a lot of work time is required because the target must be collimated and measured. In this study, the data of the mines were acquired with the total station and the 3D laser scanner, and the mine spatial information was constructed by using the shape based registration method. By using the static scanner data of some area applying the reference point surveying result of the total station, it was possible to construct the accurate result on the wide area acquired by the mobile scanner effectively. Also, the accuracy of the constructed geospatial information was evaluated and the deviation of mean 0.083m was shown. Point cloud products constructed through the research can contribute to the efficiency improvement of mine management by enabling quantitative analysis such as visualization of mine shape, distance, area and slope, and automation of drawing creation for cross section shape.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.33 no.6
• /
• pp.1099-1110
• /
• 2023

The recognition system of autonomous driving and robot navigation performs vision work such as object recognition, tracking, and lane detection after multi-sensor fusion to improve performance. Currently, research on a deep learning model based on the fusion of a camera and a lidar sensor is being actively conducted. However, deep learning models are vulnerable to adversarial attacks through modulation of input data. Attacks on the existing multi-sensor-based autonomous driving recognition system are focused on inducing obstacle detection by lowering the confidence score of the object recognition model.However, there is a limitation that an attack is possible only in the target model. In the case of attacks on the sensor fusion stage, errors in vision work after fusion can be cascaded, and this risk needs to be considered. In addition, an attack on LIDAR's point cloud data, which is difficult to judge visually, makes it difficult to determine whether it is an attack. In this study, image scaling-based camera-lidar We propose an attack method that reduces the accuracy of LCCNet, a fusion model (camera-LiDAR calibration model). The proposed method is to perform a scaling attack on the point of the input lidar. As a result of conducting an attack performance experiment by size with a scaling algorithm, an average of more than 77% of fusion errors were caused.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.6
• /
• pp.507-513
• /
• 2021

GPR (Ground Penetrating RADAR) is a sensor that inspects the pavement state of roads, sinkholes, and underground pipes. It is widely used in road management. MMS (Mobile Mapping System) creates a detailed and accurate road map of the road surface and its surroundings. If both types of data are built in the same area, it is efficient to construct both ground and underground spatial information at the same time. In addition, since it is possible to grasp the road and important facilities around the road, the location of underground pipelines, etc. without special technology, an intuitive understanding of the site is also possible, which is a useful tool in managing the road or facilities. However, overseas equipment to which this latest technology is applied is expensive and does not fit the domestic situation. LiDAR (Light Detection And Raging) and GNSS/INS (Global Navigation Satellite System / Inertial Navigation System) were synchronized in order to replace overseas developed equipment and to secure original technology to develop domestic equipment in the future, and GPR data was also synchronized to the same GNSS/INS. We developed software that performs georeferencing using the location and attitude information from GNSS/INS at the time of acquiring synchronized GPR data. The experiments were conducted on the road site by dividing the open sky and the non-open sky. The road and surrounding facilities on the ground could be easily checked through the 3D point cloud data acquired through LiDAR. Georeferenced GPR data could also be viewed with a 3D viewer along with point cloud data, and the location of underground facilities could be easily and quickly confirmed through GPR data.

• Journal of the Korean Geotechnical Society
• /
• v.37 no.11
• /
• pp.107-118
• /
• 2021

Scan-to-BIM can be precisely mod eled by measuring structures with Light Detection And Ranging (LiDAR) and build ing a 3D BIM (Building Information Modeling) model based on it, but has a limitation in that it consumes a lot of manpower, time, and cost. To overcome these limitations, studies are being conducted to perform semantic segmentation of 3D point cloud data applying deep learning algorithms, but studies on how segmentation result changes depending on learning data are insufficient. In this study, a parametric study was conducted to determine how the size and track type of railroad tunnels constituting learning data affect the semantic segmentation of railroad tunnels through deep learning. As a result of the parametric study, the similar size of the tunnels used for learning and testing, the higher segmentation accuracy, and the better results when learning through a double-track tunnel than a single-line tunnel. In addition, when the training data is composed of two or more tunnels, overall accuracy (OA) and mean intersection over union (MIoU) increased by 10% to 50%, it has been confirmed that various configurations of learning data can contribute to efficient learning.

• Journal of Broadcast Engineering
• /
• v.27 no.1
• /
• pp.44-55
• /
• 2022

In order to automatically provide information on fruits in agricultural related broadcasting contents, instance image segmentation of target fruits is required. In addition, the information on the 3D pose of the corresponding fruit may be meaningfully used. This paper represents research that provides information about tomatoes in video content. A large amount of data is required to learn the instance segmentation, but it is difficult to obtain sufficient training data. Therefore, the training data is generated through a data augmentation technique based on a small amount of real images. Compared to the result using only the real images, it is shown that the detection performance is improved as a result of learning through the synthesized image created by separating the foreground and background. As a result of learning augmented images using images created using conventional image pre-processing techniques, it was shown that higher performance was obtained than synthetic images in which foreground and background were separated. To estimate the pose from the result of object detection, a point cloud was obtained using an RGB-D camera. Then, cylinder fitting based on least square minimization is performed, and the tomato pose is estimated through the axial direction of the cylinder. We show that the results of detection, instance image segmentation, and cylinder fitting of a target object effectively through various experiments.

• Korean Journal of Remote Sensing
• /
• v.35 no.5_1
• /
• pp.705-714
• /
• 2019

Precision positioning is necessary for various use of high-resolution UAV images. Basically, GCP is used for this purpose, but in case of emergency situations or difficulty in selecting GCPs, the data shall be obtained without GCPs. This study proposed a method of improving positional accuracy for x, y coordinate of UAV based 3 dimensional point cloud data generated without GCPs. Road vector file by the public data (Open Data Portal) was used as reference data for improving location accuracy. The geometric correction of the 2 dimensional ortho-mosaic image was first performed and the transform matrix produced in this process was adopted to apply to the 3 dimensional point cloud data. The straight distance difference of 34.54 m before the correction was reduced to 1.21 m after the correction. By confirming that it is possible to improve the location accuracy of UAV images acquired without GCPs, it is expected to expand the scope of use of 3 dimensional spatial objects generated from point cloud by enabling connection and compatibility with other spatial information data.

• Journal of the Korean Association of Geographic Information Studies
• /
• v.26 no.4
• /
• pp.237-250
• /
• 2023

Orthophotos and DEMs were generated by UAV-based aerial photogrammetry and an attempt was made to apply them to detailed investigations for the production of flood traces. The cultivated area located in Goa-eup, Gumi, where the embankment collapsed and inundated inundation occurred due to the impact of 6th Typhoon Sanba in 2012, was selected as rhe target area. To obtain optimal accuracy of UAV photogrammetry performance, the UAV images were taken under the optimal placement of 19 GCPs and then point cloud, DEM, and orthoimages were generated through image processing using Pix4Dmapper software. After applying CloudCompare's CSF Filtering to separate the point cloud into ground elements and non-ground elements, a finally corrected DEM was created using only non-ground elements in GRASS GIS software. The flood level and flood depth data extracted from the final generated DEM were compared and presented with the flood level and flood depth data from existing data as of 2012 provided through the public data portal site of the Korea Land and Geospatial Informatix Corporation(LX).