• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.24 no.2
• /
• pp.217-226
• /
• 2006

Extraction and tracking objects are fundamental and important steps of the digital image processing and computer vision. Many algorithms about extracting and tracking objects have been developed. In this research, a method is suggested for tracking a moving object using a pair of CCD cameras and calculating the coordinate of the moving object. A 1/100 miniature of soccer field was made to apply the developed algorithms. After candidates were selected from the acquired images using the RGB value of a moving object (soccer ball), the object was extracted using its size (MBR size) among the candidates. And then, image coordinates of a moving object are obtained. The real-time position of a moving object is tracked in the boundary of the expected motion, which is determined by centering the moving object. The 3D position of a moving object can be obtained by conducting the relative orientation, absolute orientation, and space intersection of a pair of the CCD camera image.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.25 no.5
• /
• pp.383-392
• /
• 2007

This study focuses on co-registration and band selection, which are one of the pre-processing steps to apply the change detection technique using hyperspectral images. We carried out automatic co-registration by using the SIFT algorithm which performance was already established in the computer vision fields, and selected the bands fur change detection by estimating the noise of image through the PIFs reflecting the radiometric consistency. The EM algorithm was also applied to select the band objectively. Hyperion images were used for the proposed techniques, and non-calibrated bands and striping noises contained in Hyperion image were removed. Throughout the results, we could develop the reliable co-registration procedure which coincided with accuracy within 0.2 pixels (RMSE) for change detection, and verified that band selection depending on the visual inspection could be objective by extracting the PIFs.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.35 no.2
• /
• pp.75-80
• /
• 2017

Recently, there has been growing interest in spatial data that combines information and communication technology with smart cities. The high-precision LiDAR (Light Dectection and Ranging) equipment is mainly used to collect three-dimensional spatial data, and the acquired data is also used to model geographic features and to manage plant construction and cultural heritages which require precision. The LiDAR equipment can collect precise data, but also has limitations because they are expensive and take long time to collect data. On the other hand, in the field of computer vision, research is being conducted on the methods of acquiring image data and performing 3D reconstruction based on image data without expensive equipment. Thus, precise 3D spatial data can be constructed efficiently by collecting and processing image data using CCTVs which are installed as infrastructure facilities in smart cities. However, this method can have an accuracy problem compared to the existing equipment. In this study, experiments were conducted and the results were analyzed to increase the number of extracted matching points by applying the feature-based method and the area-based method in order to improve the precision of 3D spatial data built with image data acquired from stereo CCTVs. For techniques to extract matching points, SIFT algorithm and PATCH algorithm were used. If precise 3D reconstruction is possible using the image data from stereo CCTVs, it will be possible to collect 3D spatial data with low-cost equipment and to collect and build data in real time because image data can be easily acquired through the Web from smart-phones and drones.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.5
• /
• pp.668-674
• /
• 2022

Machine learning techniques using visual data have high usability in fields of industry and service such as scene recognition, fault detection, security and user analysis. Among these, user analysis through the videos from CCTV is one of the practical way of using vision data. Also, many studies about lightweight artificial neural network have been published to increase high usability for mobile and embedded environment so far. In this study, we propose the network combining the object detection and classification for mobile graphic processing unit. This network detects pedestrian and face, classifies age and gender from detected face. Proposed network is constructed based on MobileNet, YOLOv2 and skip connection. Both detection and classification models are trained individually and combined as 2-stage structure. Also, attention mechanism is used to improve detection and classification ability. Nvidia Jetson Nano is used to run and evaluate the proposed system.

• Geophysics and Geophysical Exploration
• /
• v.26 no.4
• /
• pp.211-228
• /
• 2023

Ground-penetrating radar (GPR) surveys are commonly used to monitor embankments, which is a nondestructive geophysical method. The results of GPR surveys can be complex, depending on the situation, and data processing and interpretation are subject to expert experiences, potentially resulting in false detection. Additionally, this process is time-intensive. Consequently, various studies have been undertaken to detect cavities in GPR survey data using deep learning methods. Deep-learning-based approaches require abundant data for training, but GPR field survey data are often scarce due to cost and other factors constaining field studies. Therefore, in this study, a deep- learning-based model was developed for embankment GPR survey cavity detection using data augmentation strategies. A dataset was constructed by collecting survey data over several years from the same embankment. A you look only once (YOLO) model, commonly used in computer vision for object detection, was employed for this purpose. By comparing and analyzing various strategies, the optimal data augmentation approach was determined. After initial model development, a stepwise process was employed, including box clustering, transfer learning, self-ensemble, and model ensemble techniques, to enhance the final model performance. The model performance was evaluated, with the results demonstrating its effectiveness in detecting cavities in embankment GPR survey data.

• Journal of Korea Water Resources Association
• /
• v.41 no.7
• /
• pp.701-709
• /
• 2008

Surface Image Velocimetry (SIV) is an instrument to measure water surface velocity by using image processing techniques. It gives us one of the easiest ways to measure water velocity. However, since it requires a set of plane survey data to estimate the velocity, it may give us some kind of misconcept that its usage would be difficult or cumbersome in spite of its handiness. If it has a feature that can estimate the plane survey data easily, it may be treated as like one of the conventional propeller velocimetries and its applicability would be improved so high. The present study is to propose a method to estimate the plane geometry of the physical coordinate with a calibrated camera. With the feature we can half-automatize the estimating procedure for the whole water velocity field. Photogrammetric technique to calculate the plane coordinates of the reference points with a calibrated camera was studied, which has originally studied for long time in the field of computer vision. By applying this technique to SIV, it is possible to estimate the location of reference coordinates for projective transform without plane survey. With this procedure the cumbersome plane survey for the reference points is omitted. One example application of the developed method showed fairly good results with insignificant errors.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.19 no.11
• /
• pp.594-603
• /
• 2018

Recently, according to the development of measurement techniques, it has become possible to take complicated and time-consuming field measurements in a simple and convenient manner. In this background, this study estimated the tension of cables under ambient vibration using minimized measurement and signal processing. The VBDM using video-only by low-cost equipment was used as a minimized measurement. An estimation of the natural frequency using the mirror frequency concept was also proposed to solve the shortage of frequency band in this case. Furthermore, the FDD method was adopted for a natural frequency estimation in the ambient vibration related to field application. Experimental studies using a cable-stayed bridge model were carried out to examine the properties of the mirror frequency and the applicability of FDD with the proposed minimized system. The results showed that FDD for ambient vibration also works properly in an estimation of the natural frequency using the minimized system. In addition, the mirror frequency concept can allow a high natural frequency estimation even in a distorted signal by low-speed recording, which can overcome the limit of the minimized system. Overall, the proposed minimized system can be effective for the tension estimations of a cable under ambient vibration.

• Journal of Intelligence and Information Systems
• /
• v.28 no.1
• /
• pp.89-106
• /
• 2022

Recently, the amount of video data collected from smartphones, CCTVs, black boxes, and high-definition cameras has increased rapidly. According to the increasing video data, the requirements for analysis and utilization are increasing. Due to the lack of skilled manpower to analyze videos in many industries, machine learning and artificial intelligence are actively used to assist manpower. In this situation, the demand for various computer vision technologies such as object detection and tracking, action detection, emotion detection, and Re-ID also increased rapidly. However, the object detection and tracking technology has many difficulties that degrade performance, such as re-appearance after the object's departure from the video recording location, and occlusion. Accordingly, action and emotion detection models based on object detection and tracking models also have difficulties in extracting data for each object. In addition, deep learning architectures consist of various models suffer from performance degradation due to bottlenects and lack of optimization. In this study, we propose an video analysis system consists of YOLOv5 based DeepSORT object tracking model, SlowFast based action recognition model, Torchreid based Re-ID model, and AWS Rekognition which is emotion recognition service. Proposed model uses single-linkage hierarchical clustering based Re-ID and some processing method which maximize hardware throughput. It has higher accuracy than the performance of the re-identification model using simple metrics, near real-time processing performance, and prevents tracking failure due to object departure and re-emergence, occlusion, etc. By continuously linking the action and facial emotion detection results of each object to the same object, it is possible to efficiently analyze videos. The re-identification model extracts a feature vector from the bounding box of object image detected by the object tracking model for each frame, and applies the single-linkage hierarchical clustering from the past frame using the extracted feature vectors to identify the same object that failed to track. Through the above process, it is possible to re-track the same object that has failed to tracking in the case of re-appearance or occlusion after leaving the video location. As a result, action and facial emotion detection results of the newly recognized object due to the tracking fails can be linked to those of the object that appeared in the past. On the other hand, as a way to improve processing performance, we introduce Bounding Box Queue by Object and Feature Queue method that can reduce RAM memory requirements while maximizing GPU memory throughput. Also we introduce the IoF(Intersection over Face) algorithm that allows facial emotion recognized through AWS Rekognition to be linked with object tracking information. The academic significance of this study is that the two-stage re-identification model can have real-time performance even in a high-cost environment that performs action and facial emotion detection according to processing techniques without reducing the accuracy by using simple metrics to achieve real-time performance. The practical implication of this study is that in various industrial fields that require action and facial emotion detection but have many difficulties due to the fails in object tracking can analyze videos effectively through proposed model. Proposed model which has high accuracy of retrace and processing performance can be used in various fields such as intelligent monitoring, observation services and behavioral or psychological analysis services where the integration of tracking information and extracted metadata creates greate industrial and business value. In the future, in order to measure the object tracking performance more precisely, there is a need to conduct an experiment using the MOT Challenge dataset, which is data used by many international conferences. We will investigate the problem that the IoF algorithm cannot solve to develop an additional complementary algorithm. In addition, we plan to conduct additional research to apply this model to various fields' dataset related to intelligent video analysis.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.32 no.5
• /
• pp.495-503
• /
• 2014

As the number of vehicles has been sharply increases, the significance of safety and effective operation issues in the parking lot is being emphasized, which takes a part of the transportation system. Recently, there have been several studies for the parking management by detecting moving object, however, recognizing numbers of fast-moving vehicles simultaneously in the picture is still a challenging problem. The parking lot in public area, or large-sized buildings has clear parking section, whereas the sensor system is configured to monitor a plurality of parking spaces. Therefore, by considering those parking lots, we suggested to develop the real-time parking availability information system by applying the real-time image processing techniques. with the help of template matching. Following the study, we wanted to provide the alternative method for parking management system through the reference template makers by recognizing movements of parked vehicles with the size and shape, regardless of direct detecting of driving movements. In addition, we evaluated the applicability and performances of the information system, presented in this study, and implemented a prototype system to simulate the parking statuses of each floor. In fat, it was possible to manage and analyze statistics about the total number of parking spaces and the number of vehicles parked through real-time video flames. We expected that the result of the study will be advanced, following the user-friendliness and cost reduction in operating parking management system and giving information by efficient analysis of parking situation.

• Journal of IKEEE
• /
• v.9 no.1 s.16
• /
• pp.7-18
• /
• 2005

A key component of an autonomous mobile robot is to localize itself accurately and build a map of the environment simultaneously. In this paper, we propose a vision-based mobile robot localization and mapping algorithm using scale-invariant features. A camera with fisheye lens facing toward to ceiling is attached to the robot to acquire high-level features with scale invariance. These features are used in map building and localization process. As pre-processing, input images from fisheye lens are calibrated to remove radial distortion then labeling and convex hull techniques are used to segment ceiling region from wall region. At initial map building process, features are calculated for segmented regions and stored in map database. Features are continuously calculated from sequential input images and matched against existing map until map building process is finished. If features are not matched, they are added to the existing map. Localization is done simultaneously with feature matching at map building process. Localization. is performed when features are matched with existing map and map building database is updated at same time. The proposed method can perform a map building in 2 minutes on $50m^2$ area. The positioning accuracy is ${\pm}13cm$, the average error on robot angle with the positioning is ${\pm}3$ degree.