• Korean Journal of Remote Sensing
• /
• v.38 no.5_3
• /
• pp.939-951
• /
• 2022

The most important factor in designing autonomous driving systems is to recognize the exact location of the vehicle within the surrounding environment. To date, various sensors and navigation systems have been used for autonomous driving systems; however, all have limitations. Therefore, the need for high-definition (HD) maps that provide high-precision infrastructure information for safe and convenient autonomous driving is increasing. HD maps are drawn using three-dimensional point cloud data acquired through a mobile mapping system (MMS). However, this process requires manual work due to the large numbers of points and drawing layers, increasing the cost and effort associated with HD mapping. The objective of this study was to improve the efficiency of HD mapping by segmenting semantic information in an MMS point cloud into six classes: roads, curbs, sidewalks, medians, lanes, and other elements. Segmentation was performed using various machine learning techniques including random forest (RF), support vector machine (SVM), k-nearest neighbor (KNN), and gradient-boosting machine (GBM), and 11 variables including geometry, color, intensity, and other road design features. MMS point cloud data for a 130-m section of a five-lane road near Minam Station in Busan, were used to evaluate the segmentation models; the average F1 scores of the models were 95.43% for RF, 92.1% for SVM, 91.05% for GBM, and 82.63% for KNN. The RF model showed the best segmentation performance, with F1 scores of 99.3%, 95.5%, 94.5%, 93.5%, and 90.1% for roads, sidewalks, curbs, medians, and lanes, respectively. The variable importance results of the RF model showed high mean decrease accuracy and mean decrease gini for XY dist. and Z dist. variables related to road design, respectively. Thus, variables related to road design contributed significantly to the segmentation of semantic information. The results of this study demonstrate the applicability of segmentation of MMS point cloud data based on machine learning, and will help to reduce the cost and effort associated with HD mapping.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.14 no.6
• /
• pp.69-76
• /
• 2015

Although traffic information has been actively analyzed using big data, it has not been used as much with the consideration of driver characteristics. Among the various types of information, road signs can directly affect the driver. Road signs must provide the optimal information that enables drivers to reach their destinations with ease as well as information suitable for navigation systems. However, present road sign rules provide standardized information, regardless of the road type or size. This study suggests a method for providing road information that will help drivers determine their behavior. First, the minimum character size that can be used on a road sign for each design speed was obtained with respect to the visibility and decipherability of a road sign. Instead of conventional diagram-based direction guidance, a scenario using split-based direction guidance was created. To verify the effectiveness of the provided information, a three-dimensional simulated road environment was constructed, and a driving simulator was used for the test. At a simple plane intersection, the driver was not greatly influenced by directional guidance, but at a complex, three-dimensional intersection, the driver preferred summary-based directional guidance, which is instinctive guidance, over diagram-based guidance. On the basis of the test results, a secondary verification test that applied split-based guidance at a three-dimensional intersection confirmed that the driver had no problems in making decisions.

• Journal of the Institute of Convergence Signal Processing
• /
• v.10 no.3
• /
• pp.164-169
• /
• 2009

This paper proposes a road detection method using BP(Back-Propagation) neural network based on texture information of the each candidate road region segmented for satellite images. To segment the candidate road regions, the histogram-based binarization method proposed by N.Otsu is firstly performed and the neighboring regions surrounding road regions are then removed. And after extracting the principal color using the histogram of the segmented foreground, the candidate road regions are classified into the regions within ${\pm}25$ of the principal color. Finally, the road regions are segmented using BP neural network based on texture information of the candidate regions. The texture information in this paper is calculated using co-occurrence matrix and is used as an input data of the BP neural network. The proposed method is based on the fact that the road has the constant intensity and shape. The experiment demonstrated the validity of the proposed method and showed 90% detection accuracy for the various images.

• 한국도로학회:학술대회논문집
• /
• 2010.09a
• /
• pp.229-234
• /
• 2010

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.20 no.5
• /
• pp.113-129
• /
• 2021

Recently, traffic congestion in the city is continuously increasing due to the expansion of the living area centered in the metropolitan area and the concentration of population in large cities. New road construction has become impossible due to the increase in land prices in downtown areas and limited sites, and the importance of efficient data-based road operation is increasingly emerging. For efficient road operation, it is essential to classify appropriate scenarios according to changes in traffic conditions and to operate optimal signals for each scenario. In this study, the Dynamic Time Warping model for cluster analysis of time series data was applied to traffic volume and speed data collected at continuous intersections for optimal scenario classification. We propose a methodology for composing an optimal signal operation scenario by analyzing the characteristics of the scenarios for each data used for classification.

• Journal of the Korean Society of Surveying, Geodesy, Photogrammetry and Cartography
• /
• v.36 no.6
• /
• pp.535-544
• /
• 2018

Autonomous driving can be limited by only using sensors if the sensor is blocked by sudden changes in surrounding environments or large features such as heavy vehicles. In order to overcome the limitations, the precise road-map has been used additionally. This study was conducted to segment and classify road objects using 3D point cloud data acquired by terrestrial mobile mapping system provided by National Geographic Information Institute. For this study, the original 3D point cloud data were pre-processed and a filtering technique was selected to separate the ground and non-ground points. In addition, the road objects corresponding to the lanes, the street lights, the safety fences were initially segmented, and then the objects were classified using the support vector machine which is a kind of machine learning. For the training data for supervised classification, only the geometric elements and the height information using the eigenvalues extracted from the road objects were used. The overall accuracy of the classification results was 87% and the kappa coefficient was 0.795. It is expected that classification accuracy will be increased if various classification items are added not only geometric elements for classifying road objects in the future.

• International Journal of Highway Engineering
• /
• v.7 no.4 s.26
• /
• pp.211-219
• /
• 2005

A selection of the analysis section for traffic accidents is used to analyze definitely the cause of accidents sorting similar accidents by a group and to raise the effect of improvement projects deciding the priority of accidents. In the existing method, an uniformly dividing method based on road mileages has been used, which has no consideration for similarities among accidents. Consequently, in recent, a slider-length method considering accident types rather than road mileages is widely used. In this study, using K-mean method, a non-hierarchical grouping technique used in the Cluster Analysis ai a applicatory method for the slider length method, a method classifies accidents that occurred the most nearby mileages into one group is proposed. To verify the proposed method, a comparison between the f-mean method and the dividing method at regular intervals on the data of a total of 25.6km lengths along Kyung-bu freeway in Pusan direction was made so that the K-mean method was proved to an effective method considering the similarities and adjacencies of accidents.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2010.07a
• /
• pp.113-116
• /
• 2010

본 논문은 보행중인 시각장애인에 장착된 카메라로부터 획득한 영상에서 보도와 차도 영역을 구분하기 위한 영역분할 기법과 질감 특징추출 기법에 대해 제안한다. 영상내의 허프 변환을 이용한 라인검출을 통해 도로 경계선을 검출하고, 분할된 영역을 원근에 따라 3 레벨로 구분하여 질감 특징성분을 추출함으로써 보도와 차도영역을 분리한다. 보도블럭이 가지는 복잡하고 다양한 특성의 패턴과 차도의 균일한 질감을 가진 영역의 특성을 비교하기 위하여 회전에 강건한 LBP, GLCM 질감 특징성분들을 이용함으로써 두 영역을 구분하였다. 제안된 방법은 낮과 밤 영상에 대해 실험한 결과 조도의 변화에 강건하게 영역을 분리할 수 있었고, 또한 보행자와 장애물이 많은 영상에서도 회전이나 폐색에 관계없이 영역 분리가 가능함을 검증하였다.

• Journal of the Korean Institute of Illuminating and Electrical Installation Engineers
• /
• v.22 no.1
• /
• pp.1-6
• /
• 2008

It is important to grasp the accurate lighting level for appropriate maintenance of the road lighting equipment. It is developed 2 division illuminance measurement systems. This system has a small size and no need to block the way and achieves the speedy measurement and high accuracy data in automatically measurement. As a result, this system can save expense and time to measure roadway lighting.

• The Journal of The Korea Institute of Intelligent Transport Systems
• /
• v.20 no.1
• /
• pp.70-85
• /
• 2021

Deep learning methodology, which has been actively studied in recent years, has improved the performance of artificial intelligence. Accordingly, systems utilizing deep learning have been proposed in various industries. In traffic systems, spatio-temporal graph modeling using GNN was found to be effective in predicting traffic speed. Still, it has a disadvantage that the model is trained inefficiently due to the memory bottleneck. Therefore, in this study, the road network is clustered through the graph clustering algorithm to reduce memory bottlenecks and simultaneously achieve superior performance. In order to verify the proposed method, the similarity of road speed distribution was measured using Jensen-Shannon divergence based on the analysis result of Incheon UTIC data. Then, the road network was clustered by spectrum clustering based on the measured similarity. As a result of the experiments, it was found that when the road network was divided into seven networks, the memory bottleneck was alleviated while recording the best performance compared to the baselines with MAE of 5.52km/h.