• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.17 no.1
• /
• pp.67-75
• /
• 2024

Crosswalks play an important role for the safe movement of pedestrians in a complex urban environment. However, for the visually impaired, crosswalks can be a big risk factor. Although assistive tools such as braille blocks and acoustic traffic lights exist for safe walking, poor management can sometimes act as a hindrance to safety. This paper proposes a method to improve accuracy in a deep learning-based real-time crosswalk detection model that can be used in applications for pedestrian assistance for the disabled at the beginning. The image was binarized by utilizing the characteristic that the white line of the crosswalk image contrasts with the road surface, and through this, the crosswalk could be better recognized and the location of the crosswalk could be more accurately identified by using two models that learned the whole and the middle part of the crosswalk, respectively. In addition, it was intended to increase accuracy by creating a boundary box that recognizes crosswalks in two stages: whole and part. Through this method, additional frames that the detection model did not detect in RGB image learning from the crosswalk image could be detected.

• Journal of Institute of Control, Robotics and Systems
• /
• v.22 no.9
• /
• pp.773-777
• /
• 2016

In this paper, a novel multi-object tracking method to track an unknown number of objects is proposed. To handle multiple object states and uncertain observations efficiently, a probability hypothesis density (PHD) filter is adopted and modified. The PHD filter is capable of reducing false positives, managing object appearances and disappearances, and estimating the multiple object trajectories in a unified framework. Although the PHD filter is robust in cluttered environments, it is vulnerable to false negatives. For this reason, we propose to exploit local observations in an RFS of the observation model. Each local observation is generated by using an online trained object detector. The main purpose of the local observation is to deal with false negatives in the PHD filtering procedure. The experimental results demonstrated that the proposed method robustly tracked multiple objects under practical situations.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.36 no.5
• /
• pp.319-333
• /
• 2018

As human beings spend more time indoors, and with the growing complexity of indoor spaces, more focus is given to indoor spatial applications and services. 3D topological networks are used for various spatial applications that involve navigation indoors such as emergency evacuation, indoor positioning, and visualization. Manually generating indoor network data is impractical and prone to errors, yet current methods in automation need expensive sensors or datasets that are difficult and expensive to obtain and process. In this research, a methodology for semi-automatically generating a 3D indoor topological model based on IndoorGML (Indoor Geographic Markup Language) is proposed. The concept of Shooting Point is defined to accommodate the usage of omnidirectional images in generating IndoorGML data. Omnidirectional images were captured at selected Shooting Points in the building using a fisheye camera lens and rotator and indoor spaces are then identified using image processing implemented in Python. Relative positions of spaces obtained from CAD (Computer-Assisted Drawing) were used to generate 3D node-relation graphs representing adjacency, connectivity, and accessibility in the study area. Subspacing is performed to more accurately depict large indoor spaces and actual pedestrian movement. Since the images provide very realistic visualization, the topological relationships were used to link them to produce an indoor virtual tour.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.11 no.9
• /
• pp.923-928
• /
• 2016

The various technologies related to the monitoring human health in real-time for the emergency situations are developing these days. Mostly the human pulse is used for measuring as the vital signs so far, but the EEG became a major research trend now. However, there are some problems measuring and sending EEG signals of all the people walking down the street to the dedicated server. Especially, there are some restrictions for collecting and sending EEG signals in 2-dimensional space in real-time. Therefore, I suggests an efficient network model using 3-dimensional space of drones to avoid the restrictions. The models are designed, simulated, and evaluated with the Opnet simulator.

• Journal of Korea Multimedia Society
• /
• v.22 no.12
• /
• pp.1457-1465
• /
• 2019

3D convolution is to stack multiple consecutive frames to form a cube, and then apply the 3D convolution kernel in the cube. In this structure, each feature map of the convolutional layer is connected to multiple adjacent sequential frames in the previous layer, thus capturing the motion information. However, due to the changes of pedestrian posture, motion and position, the convolution at the same place is inappropriate, and when the 3D convolution kernel is convoluted in the time domain, only time domain features of three consecutive frames can be extracted, which is not a good enough to get action information. This paper proposes an action recognition method based on feature fusion of 3D convolutional neural network. Based on the VGG16 network model, sending a pre-acquired optical flow image for learning, then get the time domain features, and then the feature of the time domain is extracted from the features extracted by the 3D convolutional neural network. Finally, the behavior classification is done by the SVM classifier.

• International Journal of Highway Engineering
• /
• v.15 no.3
• /
• pp.151-157
• /
• 2013

PURPOSES : This study aims to suggest the optimal spacing between speed humps which is placed at traffic calming areas including pedestrian priority zones, school zones, and residential areas. METHODS: The study measured the operating speed of vehicles passing through two successive humps by using laser gun in 0.2 seconds interval, and analysed the basic statistical characteristics of speeds data to have an insight on the relationship between spacing and speed. Assumption was made to derive the maximum spacing within which two successive humps influence each other. RESULTS: The statistically significant model explaining the relationship between spacing and 85th percentile speed of vehicles was derived as well as the maximum spacing maintained in order to take the benefits of successive installation of humps. CONCLUSIONS: Spacing of 20 meters was suggested to achieve the widely accepted target speed of 30 km/h in traffic calming zone, and spacing of 70 meters was suggested as a maximum spacing. The comparison across the studies were made and empirical reasoning the difference of results between studies was discussed as well as the future studies.

• Wind and Structures
• /
• v.5 no.2_3_4
• /
• pp.227-244
• /
• 2002

Recently, the prediction of wind environment around a building using Computational Fluid Dynamics (CFD) technique comes to be carried out at the practical design stage. However, there have been very few studies which examined the accuracy of CFD prediction of flow around a high-rise building including the velocity distribution at pedestrian level. The working group for CFD prediction of wind environment around building, which consists of researchers from several universities and private companies, was organized in the Architectural Institute of Japan (AIJ) considering such a background. At the first stage of the project, the working group planned to carry out the cross comparison of CFD results of flow around a high rise building by various numerical methods, in order to clarify the major factors which affect prediction accuracy. This paper presents the results of this comparison.

• KIEAE Journal
• /
• v.10 no.2
• /
• pp.63-69
• /
• 2010

Diverse problems in wind environment has occurred through rapid urbanization and growth of high-rise building numbers, This study aims to propose the CFD (Computational Fluid Dynamics) simulation method and evaluation standard of wind environment in site planning of high rise apartment housing. The CFD simulation method proposed in this study is not existing detail simulation, but it is the method that a designer can correct and develop the design through immediate evaluation of design options in concept design phase. Therefore, the proposed CFD simulation method of wind environment in this study uses the BIM based CFD tool in which the 3D model in concept design phase can be used as for the CFD simulation. In this paper, the study examines existing evaluation standards of comfortableness level in wind environment for pedestrian near buildings, and selects new evaluation method which is possible to apply to the proposed CFD simulation method. In addition, it is to examine calculation time-spending and appropriate mesh division method for finding CFD result which is useful to find the best design options in aspect of wind environment in concept design phase. Furthermore, it proposes the wind environment evaluation method through BIM based CFD simulation.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2015.05a
• /
• pp.126-129
• /
• 2015

Device-to-device (D2D) communications, a subset of Proximity-based Services that enables direct communication between LTE network subscribers, is gaining popularity. It is well underway to be adopted in cellular communication systems for pedestrian and connected-vehicles alike. In this paper, we briefly present our model of an Evolved Packet Core Network-assisted device discovery simulator and show the applicability of Proximity-based Services for Vehicular Ad-hoc Networks. Through the performance evaluation based on the developed simulation environment, it is shown that in case when users gather in the same vicinity, as in public transportation, LTE network data can be efficiently offloaded and multicasted through Wi-Fi for e.g. delivering traffic-related information and for the benefit of infotainment service consumers.

• Journal of IKEEE
• /
• v.23 no.4
• /
• pp.1166-1174
• /
• 2019

This paper propose and experiment advanced PersonNet, a human identification model, with advanced performance. We apply the inception layer to extract feature points, and increase the existing 32 feature points to 154. Also, we modify the CND method used by PersonNet to mitigate asymmetry, and apply weights to the feature map of pedestrian images in three parts, thereby making the features more distinct. Three databases were used for performance evaluation : CUHK01, CUHK03 and Market-1501. The experiment results showed 27-31% improvement in performance.