• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.10
• /
• pp.3955-3971
• /
• 2020

Due to the significant increase in the use of autonomous car technology, it is essential to integrate this technology with high-precision digital map data containing more precise and accurate roadway information, as compared to existing conventional map resources, to ensure the safety of self-driving operations. While existing map technologies may assist vehicles in identifying their locations via Global Positioning System, it is however difficult to update the environmental changes of roadways in these maps. Roadway vision algorithms can be useful for building autonomous vehicles that can avoid accidents and detect real-time location changes. We incorporate a hybrid architectural design that combines unsupervised classification of vision data with supervised joint fusion classification to achieve a better noise-resistant algorithm. We identify, via a deep learning approach, an intelligent hybrid fusion algorithm for fusing multimodal vision feature data for roadway classifications and characterize its improvement in accuracy over unsupervised identifications using image processing and supervised vision classifiers. We analyzed over 93,000 vision frame data collected from a test vehicle in real roadways. The performance indicators of the proposed hybrid fusion algorithm are successfully evaluated for the generation of roadway digital maps for autonomous vehicles, with a recall of 0.94, precision of 0.96, and accuracy of 0.92.

• Journal of Korea Society of Industrial Information Systems
• /
• v.19 no.2
• /
• pp.85-91
• /
• 2014

The objective of the research is to automatically recognize railway objects from MLS data in which 9 key objects including terrain, track, bed, vegetation, platform, barrier, posts, attachments, powerlines are targeted. The proposed method can be divided into two main sub-steps. First, multi-scale contextual features are extracted to take the advantage of characterizing objects of interest from different geometric levels such as point, line, volumetric and vertical profile. Second, by considering contextual interactions amongst object labels, a contextual classifier is utilized to make a prediction with local coherence. In here, the Conditional Random Field (CRF) is used to incorporate the object context. By maximizing the object label agreement in the local neighborhood, CRF model could compensate the local inconsistency prediction resulting from other local classifiers. The performance of proposed method was evaluated based on the analysis of commission and omission error and shows promising results for the practical use.

• Journal of The Korean Astronomical Society
• /
• v.45 no.2
• /
• pp.31-38
• /
• 2012

In this study we apply Support Vector Machine (SVM) to the prediction of geo-effective halo coronal mass ejections (CMEs). The SVM, which is one of machine learning algorithms, is used for the purpose of classification and regression analysis. We use halo and partial halo CMEs from January 1996 to April 2010 in the SOHO/LASCO CME Catalog for training and prediction. And we also use their associated X-ray flare classes to identify front-side halo CMEs (stronger than B1 class), and the Dst index to determine geo-effective halo CMEs (stronger than -50 nT). The combinations of the speed and the angular width of CMEs, and their associated X-ray classes are used for input features of the SVM. We make an attempt to find the best model by using cross-validation which is processed by changing kernel functions of the SVM and their parameters. As a result we obtain statistical parameters for the best model by using the speed of CME and its associated X-ray flare class as input features of the SVM: Accuracy=0.66, PODy=0.76, PODn=0.49, FAR=0.72, Bias=1.06, CSI=0.59, TSS=0.25. The performance of the statistical parameters by applying the SVM is much better than those from the simple classifications based on constant classifiers.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.15 no.6
• /
• pp.579-589
• /
• 2004

The radar signals are generally very sensitive to relative orientations between radar and target. Thus, the performance of a target recognition system significantly deteriorates as the region of aspect angles becomes broader. To address this difficulty, in this paper, we propose a method based on the multi-aspect information in order to improve the classification capability ever for a wide angular region. First, range profiles are used to extract feature vectors based on the central moments and principal component analysis(PCA). Then, a classifier with the use of multi-aspect information is applied to them, yielding an additional improvement of target recognition capability. There are two different strategies among the classifiers that can fuse the information from multi-aspect radar signals: independent methodology and dependent methodology. In this study, the performances of the two strategies are compared within the frame work of target recognition. The radar cross section(RCS) data of six aircraft models measured at compact range of Pohang University of Science and Technology are used to demonstrate and compare the performances of the two strategies.

• International Journal of Fluid Machinery and Systems
• /
• v.8 no.4
• /
• pp.264-273
• /
• 2015

This presentation describes an experimental approach for the detection of cavitation in hydraulic machines by use of ultrasonic signal analysis. Instead of using the high frequency pulses (typically 1MHz) only for transit time measurement different other signal characteristics are extracted from the individual signals and its correlation function with reference signals in order to gain knowledge of the water conditions. As the pulse repetition rate is high (typically 100Hz), statistical parameters can be extracted of the signals. The idea is to find patterns in the parameters by a classifier that can distinguish between the different water states. This classification scheme has been applied to different cavitation sections: a sphere in a water flow in circular tube at the HSLU in Lucerne, a NACA profile in a cavitation tunnel and two Francis model test turbines all at LMH in Lausanne. From the signal raw data several statistical parameters in the time and frequency domain as well as from the correlation function with reference signals have been determined. As classifiers two methods were used: neural feed forward networks and decision trees. For both classification methods realizations with lowest complexity as possible are of special interest. It is shown that two to three signal characteristics, two from the signal itself and one from the correlation function are in many cases sufficient for the detection capability. The final goal is to combine these results with operating point, vibration, acoustic emission and dynamic pressure information such that a distinction between dangerous and not dangerous cavitation is possible.

• KIPS Transactions on Software and Data Engineering
• /
• v.7 no.7
• /
• pp.259-266
• /
• 2018

In this paper, we propose a method to classify a document using a Recurrent Neural Network by extracting features considering word sense and contexts. Word2vec method is adopted to include the order and meaning of the words expressing the word in the document as a vector. Doc2vec is applied for considering the context to extract the feature of the document. RNN classifier, which includes the output of the previous node as the input of the next node, is used as the document classification method. RNN classifier presents good performance for document classification because it is suitable for sequence data among neural network classifiers. We applied GRU (Gated Recurrent Unit) model which solves the vanishing gradient problem of RNN. It also reduces computation speed. We used one Hangul document set and two English document sets for the experiments and GRU based document classifier improves performance by about 3.5% compared to CNN based document classifier.

• The Journal of the Korea Contents Association
• /
• v.10 no.1
• /
• pp.19-26
• /
• 2010

This paper proposes a classifier based on rectangular feature to detect face in real time. The goal is to realize a strong detection algorithm which satisfies both efficiency in calculation and detection performance. The proposed algorithm consists of the following three stages: Feature creation, classifier study and real time facial domain detection. Feature creation organizes a feature set with the proposed five rectangular features and calculates the feature values efficiently by using SAT (Summed-Area Tables). Classifier learning creates classifiers hierarchically by using the AdaBoost algorithm. In addition, it gets excellent detection performance by applying important face patterns repeatedly at the next level. Real time facial domain detection finds facial domains rapidly and efficiently through the classifier based on the rectangular feature that was created. Also, the recognition rate was improved by using the domain which detected a face domain as the input image and by using PCA and KNN algorithms and a Class to Class rather than the existing Point to Point technique.

• Proceedings of the Korean Association of Geographic Inforamtion Studies Conference
• /
• 2004.03a
• /
• pp.493-498
• /
• 2004

The objective of the present study is to compare various conventional and recently evolved satellite image-processing techniques and to ascertain the best possible technique that can identify and position of aquaculture farms accurately in and around the Younggwang coastal area. Several conventional techniques performed to extract such information fiom the Landsat-TM imagery do not seem to yield better information about the aquaculture farms, and lead to misclassification. The large errors between the actual and extracted aquaculture farm information are due to existence of spectral confusion and inadequate spatial resolution of the sensor. This leads to possible occurrence of mixture pixels or 'mixels' of the source of errors in the classification techniques. Understanding the confusing and mixture pixel problems requires the development of efficient methods that can enable more reliable extraction of aquaculture farm information. Thus, the more recently evolved methods such as the step-by-step partial spectral end-member extraction and linear spectral unmixing methods are introduced. The farmer one assumes that an end-member, which is often referred to as 'spectrally pure signature' of a target feature, does not appear to be a spectrally pure form, but always mix with the other features at certain proportions. The assumption of the linear spectral unmxing is that the measured reflectance of a pixel is the linear sum of the reflectance of the mixture components that make up that pixel. The classification accuracy of the step-by-step partial end-member extraction improved significantly compared to that obtained from the traditional supervised classifiers. However, this method did not distinguish the aquaculture ponds and non-aquaculture ponds within the region of the aquaculture farming areas. In contrast, the linear spectral unmixing model produced a set of fraction images for the aquaculture, water and soil. Of these, the aquaculture fraction yields good estimates about the proportion of the aquaculture farm in each pixel. The acquired proportion was compared with the values of NDVI and both are positively correlated (R$^2$ =0.91), indicating the reliability of the sub-pixel classification.ixel classification.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.43 no.5 s.311
• /
• pp.38-47
• /
• 2006

Conventional classification method uses a single classifier based on shape or motion feature. However this method exhibits a weakness if naively used since the classification performance is highly sensitive to the accuracy of moving region to be detected. The detection accuracy, in turn, depends on the condition of the image background. In this paper, we propose to resolve the drawback and thus strengthen the classification reliability by employing a Bayesian decision fusion and by optimally combining the decisions of three classifiers. The first classifier is based on shape information obtained from Fourier descriptors while the second is based on the shape information obtained from image gradients. The third classifier uses motion information. Our experimental results on the classification Performance of human and vehicle with a static camera in various directions confirm a significant improvement and indicate the superiority of the proposed decision fusion method compared to the conventional Majority Voting and Weight Average Score approaches.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.47 no.3
• /
• pp.36-41
• /
• 2010

An advanced form of Multistage Feature-based Classification Model(AMFCM), called AMFCM, is proposed in this paper. AMFCM like MFCM does not use the concatenated form of available feature vectors extracted from original data to classify each data, but uses only groups related to each feature vector to classify separately. The prpposed AMFCM improves the contribution rate used in MFCM and proposes a confusion table for each local classifier using a specific feature vector group. The confusion table for each local classifier contains accuracy information of each local classifier on each class of data. The proposed AMFCM is applied to the problem of music genre classification on a set of music data. The results demonstrate that the proposed AMFCM outperforms MFCM by 8% - 15% on average in terms of classification accuracy depending on the grouping algorithms used for local classifiers and the number of clusters.