• Korean Journal of Remote Sensing
• /
• v.37 no.3
• /
• pp.577-589
• /
• 2021

Drone mapping systems can be applied to many fields such as disaster damage investigation, environmental monitoring, and construction process monitoring. To integrate individual sensors attached to a drone, it was essential to undergo complicated procedures including time synchronization. Recently, a variety of composite sensors are released which consist of visual sensors and GPS/INS. Composite sensors integrate multi-sensory data internally, and they provide geotagged image files to users. Therefore, to use composite sensors in drone mapping systems, mapping accuracies from composite sensors should be examined. In this study, we analyzed the mapping accuracies of a composite sensor, focusing on the data acquisition area and pre-calibration effect. In the first experiment, we analyzed how mapping accuracy varies with the number of ground control points. When 2 GCPs were used for mapping, the total RMSE has been reduced by 40 cm from more than 1 m to about 60 cm. In the second experiment, we assessed mapping accuracies based on whether pre-calibration is conducted or not. Using a few ground control points showed the pre-calibration does not affect mapping accuracies. The formation of weak geometry of the image sequences has resulted that pre-calibration can be essential to decrease possible mapping errors. In the absence of ground control points, pre-calibration also can improve mapping errors. Based on this study, we expect future drone mapping systems using composite sensors will contribute to streamlining a survey and calibration process depending on the data acquisition circumstances.

• KIPS Transactions on Computer and Communication Systems
• /
• v.10 no.2
• /
• pp.29-38
• /
• 2021

The Internet including mobile networks is developing to overcoming the limitation of physical distance and providing or acquiring information from remote locations. However, the systems that use video as primary information require higher bandwidth for recognizing the situation in remote places more accurately through high-quality video as well as lower latency for faster interaction between devices and users. The emergence of the 5th generation mobile network provides features such as high bandwidth and precise location recognition that were not experienced in previous-generation technologies. In addition, the Mobile Edge Computing that minimizes network latency in the mobile network requires a change in the traditional system architecture that was composed of the existing smart device and high availability server system. However, even with 5G and MEC, since there is a limit to overcome the mobile network state fluctuations only by enhancing the network infrastructure, this study proposes a high-definition video streaming system in ultra-low latency based on the SRT protocol that provides Forward Error Correction and Fast Retransmission. The proposed system shows how to deploy software components that are developed in consideration of the nature of 5G and MEC to achieve sub-1 second latency for 4K real-time video streaming. In the last of this paper, we analyze the most significant factor in the entire video transmission process to achieve the lowest possible latency.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.1
• /
• pp.47-54
• /
• 2021

Calculation of quantity at construction sites is a factor that has a great influence on construction costs, and it is important to calculate accurate values. In this study, topographic model was created by using drone photogrammetry and drone LiDAR to estimate earthwork volume. ortho image and DSM (Digital Surface Model) were constructed for the study area by drone photogrammetry, and DEM (Digital Elevation Model) of the target area was established using drone LiDAR. And through accuracy evaluation, accuracy of each method are 0.034m, 0.35m in horizontal direction, 0.054m, 0.25m in vertical direction. Through the research, the usability of drone photogrammetry and drone LiDAR for constructing geospatial information was presented. As a result of calculating the volume of the study site, the UAV photogrammetry showed a difference of 1528.1㎥ from the GNSS (Global Navigation Satellite System) survey performance, and the 3D Laser Scanner showed difference of 160.28㎥. The difference in the volume of earthwork is due to the difference in the topographic model, and the efficiency of volume calculation by drone LiDAR could be suggested. In the future, if additional research is conducted using GNSS surveying and drone LiDAR to establish topographic model in the forest area and evaluate its usability, the efficiency of terrain model construction using drone LiDAR can be suggested.

• Ecology and Resilient Infrastructure
• /
• v.8 no.1
• /
• pp.22-31
• /
• 2021

Surface image velocimetry (SIV) is a noncontact velocimetry technique based on images. Recently, studies have been conducted on surface velocity measurements using drones to measure a wide range of velocities and discharges. However, when measuring the surface velocity using a drone, reference points must be included in the image for image correction and the calculation of the ground sample distance, which limits the flight altitude and shooting area of the drone. A technique for calculating the surface velocity that does not require reference points must be developed to maximize spatial freedom, which is the advantage of velocity measurements using drone images. In this study, a technique for calculating the surface velocity that uses only the drone position and the specifications of the drone-mounted camera, without reference points, was developed. To verify the developed surface velocity calculation technique, surface velocities were calculated at the Andong River Experiment Center and then measured with a FlowTracker. The surface velocities measured by conventional SIV using reference points and those calculated by the developed SIV method without reference points were compared. The results confirmed an average difference of approximately 4.70% from the velocity obtained by the conventional SIV and approximately 4.60% from the velocity measured by FlowTracker. The proposed technique can accurately measure the surface velocity using a drone regardless of the flight altitude, shooting area, and analysis area.

• Journal of Advanced Navigation Technology
• /
• v.25 no.1
• /
• pp.1-7
• /
• 2021

Detect and avoid (DAA) system, which is essential for the operation of UAS, detects intruding aircraft and offers the ranges of turn and climb/descent maneuver that are required to avoid the intruder. This paper uses detect and avoid alerting logic for unmanned systems (DAIDALUS) developed at NASA as a DAA algorithm. Since DAIDALUS offers ranges of avoidance maneuvers, the actual avoidance maneuver must be decided by the UAS pilot as well as the timing and method of returning to the original route. It can be readily used in real-time human-in-the-loop (HiTL) simulations where a human pilot is making the decision, but a pilot decision model is required in fast-time simulations that proceed without human pilot intervention. This paper proposes a pilot decision model that maneuvers the aircraft based on the DAIDALUS avoidance maneuver range. A series of tests were conducted using test vectors from radio technical commission for aeronautics (RTCA) minimum operational performance standards (MOPS). The alert levels differed by the types of encounters, but loss of well clear (LoWC) was avoided. This model will be useful in fast-time simulation of high-volume traffic involving UAS.

• Journal of the Korea Institute of Building Construction
• /
• v.21 no.6
• /
• pp.665-676
• /
• 2021

As the number of deteriorated buildings increases, the importance of safety diagnosis and maintenance of buildings has been rising. Existing visual investigations and building safety diagnosis objectivity and reliability are poor due to their reliance on the subjective judgment of the examiner. Therefore, this study presented the limitations of the previously conducted appearance investigation and proposed 3D Point Cloud data to increase the accuracy of existing detailed inspection data. In addition, this study conducted a calculation of an objective building safety grade using a Deep-Neural Network(DNN) structure. The DNN structure is generated using the existing detailed inspection data and precise safety diagnosis data, and the safety grade is calculated after applying the state evaluation data obtained using a 3D Point Cloud model. This proposed process was applied to 10 deteriorated buildings through the case study, and achieved a time reduction of about 50% compared to a conventional manual safety diagnosis based on the same building area. Subsequently, in this study, the accuracy of the safety grade calculation process was verified by comparing the safety grade result value with the existing value, and a DNN with a high accuracy of about 90% was constructed. This is expected to improve economic feasibility in the future by increasing the reliability of calculated safety ratings of old buildings, saving money and time compared to existing technologies.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.39 no.6
• /
• pp.541-548
• /
• 2021

Recently, multispectral cameras are being actively utilized in various application fields such as agriculture, forest management, coastal environment monitoring, and so on, particularly onboard UAV's. Resultant multispectral images are typically georeferenced primarily based on the onboard GPS (Global Positioning System) and IMU (Inertial Measurement Unit)or accurate positional information of the pixels, or could be integrated with ground control points that are directly measured on the ground. However, due to the high cost of establishing GCP's prior to the georeferencing or for inaccessible areas, it is often required to derive the positions without such reference information. This study aims to provide a means to improve the georeferencing performance of a multispectral camera images without involving such ground reference points, but instead with the simultaneously onboard high resolution RGB camera. The exterior orientation parameters of the drone camera are first estimated through the bundle adjustment, and compared with the reference values derived with the GCP's. The results showed that the incorporation of the images from a high resolution RGB camera greatly improved both the exterior orientation estimation and the georeferencing of the multispectral camera. Additionally, an evaluation performed on the direction estimation from a ground point to the sensor showed that inclusion of RGB images can reduce the angle errors more by one order.

• Journal of Internet of Things and Convergence
• /
• v.8 no.2
• /
• pp.35-40
• /
• 2022

In the agricultural environment, pesticide control requires a high risk of work and a high labor force for farmers. The effectiveness of pesticide control using unmanned aerial vehicles varies according to climate, land type, and characteristics of unmanned aerial vehicles. Therefore, an effective method for pesticide control by unmanned aerial vehicles considering the spraying conditions and environmental conditions is required. In this paper, we propose an efficient pesticide control system based on agricultural unmanned aerial vehicles considering the application conditions and environmental information for each crop. The effectiveness of the proposed model was demonstrated by measuring the drop uniformity of pesticides according to the change in altitude and speed after attaching the sensory paper and measuring the penetration rate of the drug inside the canopy according to the change in crop growth conditions. Experiment result, the closer the height of the UAV is to the ground, the more evenly the crops are sprayed, but for safety reasons, 2m more is suitable, and on average a speed of 2m/s is most suitable for control. The proposed control system is expected to help develop intelligent services based on the use of various unmanned aerial vehicles in agricultural environments.

• Korean Journal of Remote Sensing
• /
• v.38 no.2
• /
• pp.199-213
• /
• 2022

The unsupervised domain adaptation can solve the impractical issue of repeatedly collecting high-quality training data every year for annual crop classification. This study evaluates the applicability of deep learning-based unsupervised domain adaptation models for crop classification. Three unsupervised domain adaptation models including a deep adaptation network (DAN), a deep reconstruction-classification network, and a domain adversarial neural network (DANN) are quantitatively compared via a crop classification experiment using unmanned aerial vehicle images in Hapcheon-gun and Changnyeong-gun, the major garlic and onion cultivation areas in Korea. As source baseline and target baseline models, convolutional neural networks (CNNs) are additionally applied to evaluate the classification performance of the unsupervised domain adaptation models. The three unsupervised domain adaptation models outperformed the source baseline CNN, but the different classification performances were observed depending on the degree of inconsistency between data distributions in source and target images. The classification accuracy of DAN was higher than that of the other two models when the inconsistency between source and target images was low, whereas DANN has the best classification performance when the inconsistency between source and target images was high. Therefore, the extent to which data distributions of the source and target images match should be considered to select the best unsupervised domain adaptation model to generate reliable classification results.

• Economic and Environmental Geology
• /
• v.55 no.2
• /
• pp.197-207
• /
• 2022

This study investigated spectroscopic exploration of Myeonsan formation, the titanium(Ti) ore hostrock, in Joseon supergroup based on machine learning technique. The mineral composition, Ti concentration, spectral characteristics of Myeonsan and non-Myeonsan formation of Joseon supergroup were analyzed. The Myeonsan formation contains relatively larger quantity of opaque minerals along with quartz and clay minerals. The PXRF analysis revealed that the Ti concentration of Myeosan formation is at least 10 times larger than the other formations with bi-modal distribution. The bi-modal concentration is caused by high Ti concentrated sandy layer and relatively lower Ti concentrated muddy layer. The spectral characteristics of Myeonsan formation is manifested by Fe oxides at near infrared and clay minerals at shortwave infrared bands. The Ti exploration is expected to be more effective on detection of hostrock rather than Ti ore because ilmenite does not have characteristic spectral features. The random-forest machine learning classification detected the Myeonsan fomation at 85% accuracy with overall accuracy of 97%, where spectral features of iron oxides and clay minerals played an important role. It indicates that spectral analysis can detect the Ti host rock effectively, and can contribute for UAV based remote sensing for Ti exploration.