• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.41 no.3
• /
• pp.67-77
• /
• 2004

The recent research has been focused on fusion of audio and visual features for reliable speech recognition in noisy environments. In this paper, we propose a neural network based model of robust speech recognition by integrating audio, visual, and contextual information. Bimodal Neural Network(BMNN) is a multi-layer perception of 4 layers, each of which performs a certain level of abstraction of input features. In BMNN the third layer combines audio md visual features of speech to compensate loss of audio information caused by noise. In order to improve the accuracy of speech recognition in noisy environments, we also propose a post-processing based on contextual information which are sequential patterns of words spoken by a user. Our experimental results show that our model outperforms any single mode models. Particularly, when we use the contextual information, we can obtain over 90% recognition accuracy even in noisy environments, which is a significant improvement compared with the state of art in speech recognition. Our research demonstrates that diverse sources of information need to be integrated to improve the accuracy of speech recognition particularly in noisy environments.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.45 no.5
• /
• pp.103-110
• /
• 2008

A reliable extraction of welded area is the precedent task before the detection of weld defects in industrial radiography. This paper describes an attempt to detect and extract the welded features of steel tubes from the computed radiography(CR) images. The statistical properties are first analyzed on over 160 sample radiographic images which represent either weld or non-weld area to identify the differences between them. The analysis is then proceeded by pattern classification to determine the clustering parameters. These parameters are the width, the functional match, and continuity. The observed weld image is processed line by line to calculate these parameters for each flexible moving window in line image pixel set. The local statistic line-clustering method is used as the classifier to recognize each window data as weld or non-weld cluster. The sequential procedure is to track the edge lines between two distinct regions by iterative calculation of threshold, and it results in extracting the weld feature. Our methodology is concluded to be effective after experiment with CR weld images.

• The Journal of Korean Institute of Electromagnetic Engineering and Science
• /
• v.11 no.5
• /
• pp.748-754
• /
• 2000

A novel single feed wide band CP stacked microstrip antenna using crossed slots has been designed, fabricated and measured. For the single rediating element the designed 10dB return loss bandwidth is 34.5%99.45~13.54 GHz), 3dB axial ratio bandwidth is 18.7%(11.17~13.39GHz), and 6 dB gain bandwidth is 29%(10.21~13.64GHz). For the 2$\times$2 array designed using a sequential rotation method, the 10dB return loss bandwidth is 35.9%(9.69~13.94GHz), 3dB axial ratio bandwidth is 34.6GHz (9.93~14.03GHz), and 6dB gain bandwidth is 27.4%(10.35~13.6GHz). For the fabricated 8$\times$8 array antenna, the 10dB return loss bandwidth is 27.3%(10.17~13.41GHz), 3dB axial ratio bandwidth is 27.9GHz(10.1~13.4GHz), and the radiation pattern is good agreement with theory. This antenna can be used for broadband applications for communications or broadcasting in Ku band.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.17 no.2
• /
• pp.159-167
• /
• 1992

This paper describes a feature extraction process by using a multi-layer neural network and is applied to the Korean stroke pattern for on line hand written character recognition, In the first layer the features are detected during the writing process and in the second layer the stroke specific features are extracted. A modified Masking field algorithm for direction co9nstancy has been used in this neural network and the resulting action potential of stroke specific features represents statistical distribution of the features in the on-line input stroke pattern and these results can be used in the recognition of on-line hand written Korean characters successfully.

• Journal of the Korea Society of Computer and Information
• /
• v.12 no.5
• /
• pp.39-47
• /
• 2007

In this thesis, a method of generating an adaptable Region-Of-Interest(ROI) Mask for the Region-Of-Interest coding is suggested. In the method, an ROI Mask is generated using the information of the ROI designated by a user. In the existed method of ROI coding, after scanning all the pixels in order and discriminating an ROI, an ROI Mask is generated. But, in our method, after scanning a part of pixels based on the shape pattern of an ROI and discriminating a ROI by one code block unit, an ROI Mask is generated. Moreover, from the method, a pattern number, threshold of a ROI and background threshold parameter are provided. According to the result of its comparing test with the existed methods to show the usability, it is proved that our method is superior in speed to the existed ones.

• Journal of KIISE:Databases
• /
• v.28 no.3
• /
• pp.301-314
• /
• 2001

Data mining refers to a set of techniques for discovering implicit and useful knowledge from large database. Many studies on data mining have been pursued and some of them have involved issues of temporal data mining for discovering knowledge from temporal database, such as sequential pattern, similar time sequence, cyclic and temporal association rules, etc. However, all of the works treat problems for discovering temporal pattern from data which are stamped with time points and do not consider problems for discovering knowledge from temporal interval data. For example, there are many examples of temporal interval data that it can discover useful knowledge from. These include patient histories, purchaser histories, web log, and so on. Allen introduces relationships between intervals and operators for reasoning about relations between intervals. We present a new data mining technique that can discover temporal relation rules in temporal interval data by using the Allen's theory. In this paper, we present two new algorithms for discovering algorithm for generating temporal relation rules, discovers rules from temporal interval data. This technique can discover more useful knowledge in compared with conventional data mining techniques.

• Journal of the Korean association of regional geographers
• /
• v.10 no.3
• /
• pp.667-681
• /
• 2004

This research is aimed to investigate the relationship of the effective wind velocity and the morphological change in coastal dune at Sindu-ri in Korea. Sediment flux was estimated by the measurement of elevation change along eight transects deployed in the study area from July 1999 to July 2000. The results of this study indicated that. first. based on the occurrence of morphological change and winds effective for sand movement. two distinct seasons were identified: a season of deposition and that of erosion. During the depositional season, spanning from November to April, effective winds were dominant and sand accumulation occurred mainly in foredunes and sequentially in dune plains. During the erosional season, from May through October, volume change was small and erosion or sand loss occurred mostly at the dunefoot of foredunes. Secondly, the research revealed that the sediment budget of Sindu coastal dune turned out to be surplus on the whole, but there are some regional differences. Deficit budgets were observed partly in secondary dunes. The utmost northern part of Sindu coastal dune was provided with abundant sand, whereas the central and northern parts were poorly supplied.

• Journal of KIISE:Databases
• /
• v.31 no.6
• /
• pp.699-709
• /
• 2004

XML is increasingly important in data exchange and information management. Starting point for retrieving the structure and integrating the documents efficiently is clustering the documents that have similar structure. The reason is that we can retrieve the documents more flexible and faster than the method treating the whole documents that have different structure. Therefore, in this paper, we propose the similar structure-based incremental clustering method useful for retrieving the structure of XML documents and integrating them. As a novel method, we use a clustering algorithm for transactional data that facilitates the large number of data, which is quite different from the existing methods that measure the similarity between documents, using vector. We first extract the representative structures of XML documents using sequential pattern algorithm, and then we perform the similar structure based document clustering, assuming that the document as a transaction, the representative structure of the document as the items of the transaction. In addition, we define the cluster cohesion and inter-cluster similarity, and analyze the efficiency of the Proposed method through comparing with the existing method by experiments.

• The KIPS Transactions:PartA
• /
• v.8A no.4
• /
• pp.447-454
• /
• 2001

Various caching and prefetching algorithms have been investigated to identify and effective method for improving the performance of I/O devices. A prefetching algorithm decreases the processing time of a system by reducing the number of disk accesses when an I/O is needed. This paper proposes an AMBA prefetching method that is an extended version of the OBA prefetching method. The AMBA prefetching method will prefetching blocks continuously as long as disk bandwidth is enough. In this method, though there were excessive data request rate, we would expect efficient prefetching. And in the AMBA prefetching method, to prevent the cache pollution, it limits the number of data blocks to be prefetched within the cache area. It can be implemented in a user-level File System based on a Linux Operating System. In particular, the proposed prefetching policy improves the system performance by about 30∼40% for large files that are accessed sequentially.

• Journal of KIISE
• /
• v.42 no.7
• /
• pp.834-839
• /
• 2015

Recently, wellness has become an issue related to improvements in personal health and quality of life. Data that are accumulated daily, such as meals and momentum records, in addition to body measurement information such as body weight, BMI and blood pressure have been used to analyze the personal health data of an individual. Therefore, it has become possible to prevent potential disease and to analyze dietary or exercise patterns. In terms of food and nutrition, analyses are performed to evaluate the health status of an individual using dietary data. However, it is very difficult to process the large amount of dietary data. An analysis of dietary data includes four steps, and each step contains a series of iterative tasks that are executed over a long time. This paper proposes a problem solving environment that automates dietary data analysis, and the proposed framework increases the speed with which an experiment can be conducted.