• Korean Journal of Optics and Photonics
• /
• v.27 no.5
• /
• pp.165-173
• /
• 2016

In this paper, we propose a three-dimensional (3D) scanning system based on two line lasers. This system uses two line lasers with different wavelengths as light sources. 532-nm and 630-nm line lasers can compensate for missing scan data generated by geometrical occlusion. It also can classify two laser planes by using the red and green channels. For automatic registration of scanning data, we control a stepping motor and divide the motor's rotational degree of freedom into micro-steps. To this end, we design a control printed circuit board for the laser and stepping motor, and use an image processing board. To compute a 3D point cloud, we obtain 200 and 400 images with laser lines and segment lines on the images at different degrees of rotation. The segmented lines are thinned for one-to-one matching of an image pixel with a 3D point.

• 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.

• Ecology and Resilient Infrastructure
• /
• v.7 no.2
• /
• pp.90-96
• /
• 2020

Recently, a method that applies laser scanning (LS) that acquires vegetation information such as the vegetation habitat area and the size of vegetation in a point cloud format has been proposed. When LS is used to investigate the physical shape of vegetation, it has the advantage of more accurate and rapid information acquisition. However, to examine uncertainties that may arise during measurement or post-processing, the process of adjusting the data by the actual data is necessary. Therefore, in this study, the physical structure of stems, branches, and leaves of woody vegetation in an artificially formed river channel was manually investigated. The obtained results then compared with the information acquired using the three-dimensional terrestrial laser scanning (3D TLS) method, which repeatedly scanned the target vegetation in various directions to obtain relevant information with improved precision. The analysis demonstrated a negligible difference between the measurements for the diameters of vegetation and the length of stems; however, in the case of branch length measurement, a relatively more significant difference was observed. It is because the implementation of point cloud information limits the precise differentiation between branches and leaves in the canopy area.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.32 no.4_1
• /
• pp.319-326
• /
• 2014

The accurate geo-referencing processes that apply ground control points is prerequisite for effective end use of HRSI (High-resolution satellite imagery). Since the conventional control point acquisition by human operator takes long time, demands for the automated matching to existing reference data has been increasing its popularity. Among many options of reference data, the airborne LiDAR (Light Detection And Ranging) data shows high potential due to its high spatial resolution and vertical accuracy. Additionally, it is in the form of 3-dimensional point cloud free from the relief displacement. Recently, a new matching method between LiDAR data and HRSI was proposed that is based on the image projection of whole LiDAR data into HRSI domain, however, importing and processing the large amount of LiDAR data considered as time-consuming. Therefore, we wmotivated to ere propose a local LiDAR chip generation for the HRSI geo-referencing. In the procedure, a LiDAR point cloud was rasterized into an ortho image with the digital elevation model. After then, we selected local areas, which of containing meaningful amount of edge information to create LiDAR chips of small data size. We tested the LiDAR chips for fully-automated geo-referencing with Kompsat-2 and Kompsat-3 data. Finally, the experimental results showed one-pixel level of mean accuracy.

• Korean Journal of Remote Sensing
• /
• v.28 no.5
• /
• pp.573-586
• /
• 2012

Point clouds acquired by a LIDAR(Light Detection And Ranging, also LADAR) system often contain erroneous points called outliers seeming not to be on physical surfaces, which should be carefully detected and eliminated before further processing for applications. Particularly in case of LIDAR systems employing with a Gieger-mode array detector (GmFPA) of high sensitivity, the outlier ratio is significantly high, which makes existing algorithms often fail to detect the outliers from such a data set. In this paper, we propose a method to discriminate outliers from a point cloud with high outlier ratio acquired by a GmFPA LIDAR system. The underlying assumption of this method is that a meaningful targe surface occupy at least two adjacent pixels and the ranges from these pixels are similar. We applied the proposed method to simulated LIDAR data of different point density and outlier ratio and analyzed the performance according to different thresholds and data properties. Consequently, we found that the outlier detection probabilities are about 99% in most cases. We also confirmed that the proposed method is robust to data properties and less sensitive to the thresholds. The method will be effectively utilized for on-line realtime processing and post-processing of GmFPA LIDAR data.

• Journal of Intelligence and Information Systems
• /
• v.26 no.2
• /
• pp.1-25
• /
• 2020

In this paper, we suggest an application system architecture which provides accurate, fast and efficient automatic gasometer reading function. The system captures gasometer image using mobile device camera, transmits the image to a cloud server on top of private LTE network, and analyzes the image to extract character information of device ID and gas usage amount by selective optical character recognition based on deep learning technology. In general, there are many types of character in an image and optical character recognition technology extracts all character information in an image. But some applications need to ignore non-of-interest types of character and only have to focus on some specific types of characters. For an example of the application, automatic gasometer reading system only need to extract device ID and gas usage amount character information from gasometer images to send bill to users. Non-of-interest character strings, such as device type, manufacturer, manufacturing date, specification and etc., are not valuable information to the application. Thus, the application have to analyze point of interest region and specific types of characters to extract valuable information only. We adopted CNN (Convolutional Neural Network) based object detection and CRNN (Convolutional Recurrent Neural Network) technology for selective optical character recognition which only analyze point of interest region for selective character information extraction. We build up 3 neural networks for the application system. The first is a convolutional neural network which detects point of interest region of gas usage amount and device ID information character strings, the second is another convolutional neural network which transforms spatial information of point of interest region to spatial sequential feature vectors, and the third is bi-directional long short term memory network which converts spatial sequential information to character strings using time-series analysis mapping from feature vectors to character strings. In this research, point of interest character strings are device ID and gas usage amount. Device ID consists of 12 arabic character strings and gas usage amount consists of 4 ~ 5 arabic character strings. All system components are implemented in Amazon Web Service Cloud with Intel Zeon E5-2686 v4 CPU and NVidia TESLA V100 GPU. The system architecture adopts master-lave processing structure for efficient and fast parallel processing coping with about 700,000 requests per day. Mobile device captures gasometer image and transmits to master process in AWS cloud. Master process runs on Intel Zeon CPU and pushes reading request from mobile device to an input queue with FIFO (First In First Out) structure. Slave process consists of 3 types of deep neural networks which conduct character recognition process and runs on NVidia GPU module. Slave process is always polling the input queue to get recognition request. If there are some requests from master process in the input queue, slave process converts the image in the input queue to device ID character string, gas usage amount character string and position information of the strings, returns the information to output queue, and switch to idle mode to poll the input queue. Master process gets final information form the output queue and delivers the information to the mobile device. We used total 27,120 gasometer images for training, validation and testing of 3 types of deep neural network. 22,985 images were used for training and validation, 4,135 images were used for testing. We randomly splitted 22,985 images with 8:2 ratio for training and validation respectively for each training epoch. 4,135 test image were categorized into 5 types (Normal, noise, reflex, scale and slant). Normal data is clean image data, noise means image with noise signal, relfex means image with light reflection in gasometer region, scale means images with small object size due to long-distance capturing and slant means images which is not horizontally flat. Final character string recognition accuracies for device ID and gas usage amount of normal data are 0.960 and 0.864 respectively.

• Journal of Energy Engineering
• /
• v.25 no.1
• /
• pp.1-8
• /
• 2016

When the oil is produced in the low temperature environment, wax can be accumulated in petroleum production system(pipeline, riser) and causes problems such as pipeline stucking, disturbance of the oil production. These problems can be lead to time-consuming and economic losses for flow assurance. For prediction and mitigation of wax deposition, it is necessary to measure the Wax Appearance Temperature(WAT) which is a temperature when the wax crystals start to be formed. WAT standard measurement method of transparent oil has to determine the cloud point of sample to the naked eye and cannot be applied to continuous change of the temperature. In this study, wax behavior of transparent oil samples are recorded depending on temperature using Visualized WAT Measurement System. Also, WATs of transparent oil samples are measured by image processing and compared with the result of the standard method.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.17 no.1
• /
• pp.33-40
• /
• 2022

With the recent increase in the development of military drones, they are adopted and used as the combat system of battalion level or higher. However, it is difficult to use drones that can be used in battles below the platoon level due to the current conditions for the formation of units in the Korean military. In this paper, therefore, we developed a program drones equipped with a thermal imaging camera and LiDAR sensor for reconnaissance and exploration that can be applied in battles below the platoon level. Using these drones, we studied the possibility and feasibility of drones for small-scale combats that can find hidden enemies, search for an appropriate detour through image processing and conduct reconnaissance and search for battlefields, hiding and cover-up through image processing. In addition to the purpose of using the proposed drone to search for an enemies lying in ambush in the battlefield, it can be used as a function to check the optimal movement path when a combat unit is moving, or as a function to check the optimal place for cover-up or hiding. In particular, it is possible to check another route other than the route recommended by the program because the features of the terrain can be checked from various viewpoints through 3D modeling. We verified the possiblity of flying by designing and assembling in a form of adding LiDAR and thermal imaging camera module to a drone assembled based on racing drone parts, which are open source hardware, and developed autonomous flight and search functions which can be used even by non-professional drone operators based on open source software, and then installed them to verify their feasibility.

• 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 Cadastre & Land InformatiX
• /
• v.50 no.2
• /
• pp.155-168
• /
• 2020

Open source is a key growth engine of the 4th industrial revolution, and the continuous development and use of various algorithms for image processing is expected. The purpose of this study is to examine the effectiveness of the UAS image processing open source based algorithm by comparing and analyzing the water reproduction and moving object filtering function and the time required for data processing in 3D reproduction. Five matching algorithms were compared based on recall and processing speed through the 'ANN-Benchmarks' program, and HNSW (Hierarchical Navigable Small World) matching algorithm was judged to be the best. Based on this, 108 algorithms for image processing were constructed by combining each methods of triangulation, point cloud data densification, and surface generation. In addition, the 3D reproduction and data processing time of 108 algorithms for image processing were studied for UAS (Unmanned Aerial System) images of a park adjacent to the sea, and compared and analyzed with the commercial image processing software 'Pix4D Mapper'. As a result of the study, the algorithms that are good in terms of reproducing water and filtering functions of moving objects during 3D reproduction were specified, respectively, and the algorithm with the lowest required time was selected, and the effectiveness of the algorithm was verified by comparing it with the result of 'Pix4D Mapper'.