• Korean Journal of Artificial Intelligence
• /
• v.11 no.4
• /
• pp.1-8
• /
• 2023

In this paper, we propose a black ice detection platform framework using Convolutional Neural Networks (CNNs). To overcome black ice problem, we introduce a real-time based early warning platform using CNN-based architecture, and furthermore, in order to enhance the accuracy of black ice detection, we apply a multi-scale dilation convolution feature fusion (MsDC-FF) technique. Then, we establish a specialized experimental platform by using a comprehensive dataset of thermal road black ice images for a training and evaluation purpose. Experimental results of a real-time black ice detection platform show the better performance of our proposed network model compared to conventional image segmentation models. Our proposed platform have achieved real-time segmentation of road black ice areas by deploying a road black ice area segmentation network on the edge device Jetson Nano devices. This approach in parallel using multi-scale dilated convolutions with different dilation rates had faster segmentation speeds due to its smaller model parameters. The proposed MsCD-FF Net(2) model had the fastest segmentation speed at 5.53 frame per second (FPS). Thereby encouraging safe driving for motorists and providing decision support for road surface management in the road traffic monitoring department.

• The Journal of Korea Robotics Society
• /
• v.12 no.2
• /
• pp.217-227
• /
• 2017

Crosswalk detection is an important part of the Pedestrian Protection System in autonomous vehicles. Different methods of crosswalk detection have been introduced so far using crosswalk edge features, the distance between crosswalk blocks, laser scanning, Hough Transformation, and Fourier Transformation. However, most of these methods failed to detect crosswalks accurately, when they are damaged, faded away or partly occluded. Furthermore, these methods face difficulties when applying on real road environment where there are lot of vehicles. In this paper, we solve this problem by first using a region based binarization technique and x-axis histogram to detect the candidate crosswalk areas. Then, we apply Support Vector Machine (SVM) based classification method to decide whether the candidate areas contain a crosswalk or not. Experiment results prove that our method can detect crosswalks in different environment conditions with higher recognition rate even they are faded away or partly occluded.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.1
• /
• pp.154-160
• /
• 1995

Maps are one of the most complicated types of drawings. Drawing recognition technology is not yet sophisticated enough for automated map reading. To automatically extract a road map directly form more complicated topographical maps, a very complicated algorithm is needed, simce the image generally involves such complicated patterns as symbols, characters, residential sections, rivers,etc. This paper describes a new feature extraction method based on the human optical neural field. We apply this method to extract complete set of road segments from topographical maps. The proposed method successfully extract road segments from various areas.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.11
• /
• pp.1496-1505
• /
• 1995

Maps are one of the most complicated types of drawings. Drawing recognition technology is not yet sophisticated enough for automated map reading. To automatically extract a road map dircetly from complicated topographical maps, a very sophisticated algorithm is needed, since the image generally involvfes such complicated patterns as symbols, characters, residential sections, rivers, railroads, etc. This paper proposes a new feature extraction method based on the morphology. We apply this method to extract complete set of road segments from topographical maps. The proposed method successfully extract road segments from various areas.

• Journal of the Korean housing association
• /
• v.21 no.5
• /
• pp.137-144
• /
• 2010

The purpose of this study is to clarify locational characteristics and the road systems of existing 22 old settlements in Cheongju downtown of Korea. These old settlements are classified largely into two types by geographical feature; Hill type located in ridge and Valley type in the valley from the ridge. The Hill type is further categorized into San-rock located on ridge and Gogae located on hillside, and the Valley type into closed large and small valley, that is, Daegok and Sogok by the shape and size of the valley. The road systems are classified into outer road, approaching road, inner road, and side road. In San-rock type villages, Inner road is changed side roads by new arterial road, and new ones become inner road. The outer road are almost cut off, but side roads maintain. In the case of the Gogae type, main road becomes inner road, maintain old hierarchical road, and in other cases, villages are formed around side roads derived from main road, and outer road and inner road are cut off. Among the Valley type, Daegok type maintains old hierarchical road, while for Sogok type, outer roads are cut off, and new roads are outer road.

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.79-82
• /
• 2005

Recently the necessity of road data is still being increased in industrial society, so there are many repairing and new constructions of roads at many areas. According to the development of government, city and region, the update and acquisition of road data for GIS (Geographical Information System) is very necessary. In this study, the fusion method with range data(3D Ground Coordinate System Data) and Intensity data in stand alone LiDAR data is used for road extraction and then digital image processing method is applicable. Up to date Intensity data of LiDAR is being studied. This study shows the possibility method for road extraction using Intensity data. Intensity and Range data are acquired at the same time. Therefore LiDAR does not have problems of multi-sensor data fusion method. Also the advantage of intensity data is already geocoded, same scale of real world and can make ortho-photo. Lastly, analysis of quantitative and quality is showed with extracted road image which compare with I: 1,000 digital map.

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.525-527
• /
• 2003

Recently, the spatial resolution of earth observation satellites is significantly increased to a few meters. Such high spatial resolution images definitely will provide lots of information for detail-thirsty remote sensing users. However, it is more difficult to develop automated image algorithms for automated image feature extraction and pattern recognition. In this study, we propose a two-stage procedure to extract road information from high resolution satellite images. At first stage, a watershed segmentation technique is developed to classify the image into various regions. Then, a knowledge is built for road and used to extract the road regions. In this study, we use panchromatic and multi-spectral images of the IKONOS satellite as test dataset. The experiment result shows that the proposed technique can generate suitable and meaningful road objects from high spatial resolution satellite images. Apparently, misclassified regions such as parking lots are recognized as road needed further refinement in future research.

• The Transactions of the Korea Information Processing Society
• /
• v.1 no.4
• /
• pp.452-460
• /
• 1994

Maps are one of the most complicated types of drawings. Drawing recognition technology is not yet sophisticated enough for automated map reading To automatically extract a road map directly from more complicated topographical maps, a very complicated algorithm is needed, since the image generally involves such complicated patterns as symbols, characters, residential sections, rivers, railroads, etc. This paper describes a new feature extraction method based on the human optical neural field. We apply this method to extract complete set of road segments from topographical maps. The proposed method successfully extract road segments from various areas.

• Proceedings of the KSRS Conference
• /
• 2002.10a
• /
• pp.34-39
• /
• 2002

Road information is very important for topographic mapping, transportation application, urban planning and other related application fields. Therefore, automatic detection of road networks from spatial imagery, such as aerial photos and satellite imagery can play a central role in road information acquisition. In this paper, we use least squares correlation matching alone for road center tracking and show that it works. We assumed that (bright) road centerlines would be visible in the image. We further assumed that within a same road segment, there would be only small differences in brightness values. This algorithm works by defining a template around a user-given input point, which shall lie on a road centerline, and then by matching the template against the image along the orientation of the road under consideration. Once matching succeeds, new match proceeds by shifting a matched target window further along road orientation at the target window. By repeating the process above, we obtain a series of points, which lie on a road centerline successively. A 1m resolution IKONOS images over Seoul and Daejeon were used for tests. The results showed that this algorithm could extract road centerlines in any orientation and help in fast and exact he ad-up digitization/vectorization of cartographic images.

• International Journal of Internet, Broadcasting and Communication
• /
• v.7 no.2
• /
• pp.123-129
• /
• 2015

This paper proposes an efficient method to detect direction indicators on road surfaces to support drivers in driving safely using the Simulink model. In the proposed method, the ROIs are detected using the detection method of maximally stable extremal regions (MSER), and the road indicator regions are detected using the speeded up robust features (SURF) matching method for the corresponding point matching of the detected ROIs and the road indicator templates. Experiments on various road satiations show that the processing time of about 0.32 sec per frame was required, and a detection rate of 91% was achieved.