• MALSORI
• /
• no.61
• /
• pp.87-100
• /
• 2007

A punctuation generation system which combines prosodic information with acoustic and language model information is presented. Experiments have been conducted first for the reference text transcriptions. In these experiments, prosodic information was shown to be more useful than language model information. When these information sources are combined, an F-measure of up to 0.7830 was obtained for adding punctuation to a reference transcription. This method of punctuation generation can also be applied to the 1-best output of a speech recogniser. The 1-best output is first time aligned. Based on the time alignment information, prosodic features are generated. As in the approach applied in the punctuation generation for reference transcriptions, the best sequence of punctuation marks for this 1-best output is found using the prosodic feature model and an language model trained on texts which contain punctuation marks.

• Journal of Information Processing Systems
• /
• v.18 no.3
• /
• pp.311-318
• /
• 2022

Tobacco sales enterprises often need to summarize and verify the daily sales bills, which may consume substantial manpower, and manual verification is prone to occasional errors. The use of artificial intelligence technology to realize the automatic identification and verification of such bills offers important practical significance. This study presents a novel multi-branch residual network for tobacco sales bills to improve the efficiency and accuracy of tobacco sales. First, geometric correction and edge alignment were performed on the input sales bill image. Second, the multi-branch residual network recognition model is established and trained using the preprocessed data. The comparative experimental results demonstrated that the correct recognition rate of the proposed method reached 98.84% on the China Tobacco Bill Image dataset, which is superior to that of most existing recognition methods.

• The KIPS Transactions:PartB
• /
• v.14B no.3 s.113
• /
• pp.145-152
• /
• 2007

In this paper, we propose an automatic extraction model for unknown translations and implement an unknown translation extraction system using the proposed model. The proposed model as a phrase-alignment model is incorporated with three models: a phrase-boundary model, a language model, and a translation model. Using the proposed model we implement the system for extracting unknown translations, which consists of three parts: construction of parallel corpora, alignment of Korean and English words, extraction of unknown translations. To evaluate the performance of the proposed system, we have established the reference corpus for extracting unknown translation, which comprises of 2,220 parallel sentences including about 1,500 unknown translations. Through several experiments, we have observed that the proposed model is very useful for extracting unknown translations. In the future, researches on objective evaluation and establishment of parallel corpora with good quality should be performed and studies on improving the performance of unknown translation extraction should be kept up.

• Journal of the Korea Institute of Information Security & Cryptology
• /
• v.18 no.1
• /
• pp.11-21
• /
• 2008

Biometrics-based user authentication has several advantages over traditional password-based systems for standalone authentication applications. This is also true for new authentication architectures known as crypto-biometric systems, where cryptography and biometrics are merged to achieve high security and user convenience at the same time. Recently, a cryptographic construct, called fuzzy vault, has been proposed for crypto-biometric systems. This construct aims to secure critical data(e.g., secret key) with the fingerprint data in a way that only the authorized user can access the secret by providing the valid fingerprint, and some implementations results for fingerprint have been reported. However, the previous results had some limitation of the provided security due to the limited numbers of chaff data fer hiding real fingerprint data. In this paper, we propose an approach to provide both the automatic alignment of fingerprint data and higher security by using a 3D geometric hash table. Based on the experimental results, we confirm that the proposed approach of using the 3D geometric hash table with the idea of the fuzzy vault can perform the fingerprint verification securely even with more chaff data included.

• Journal of the Institute of Electronics Engineers of Korea SD
• /
• v.42 no.11
• /
• pp.23-32
• /
• 2005

The automatic generation of circuit diagrams has been practically used in the HDL based design for decades. Nevertheless, the diagrams became too complicated for the designers to identify the signal flows in the RTL and system level designs. In this paper, we propose four techniques to enhance the roadability of the complicated diagrams. They include i) the transformation of repetitive instances and terminals into vector forms, ii) an improved loop breaking algorithm, iii) a flat tap which simplifies the two level bus ripping structure that is required for the connection of a bundle net to multiple buses, and iv) the identification of block strings, and alignment of the corresponding blocks. Towards validating the proposed techniques, the diagrams of an industrial strength design m generated. The complexity of the diagrams has been reduced by up to $90\%$ in terms of the number of wires, the aggregate wire length, and the area.

• Journal of the Korean Society of Manufacturing Technology Engineers
• /
• v.23 no.1
• /
• pp.94-101
• /
• 2014

Recently, in order to reduce the time required to replace an optical jumper cord, many researchers are using a field-installable connector and applying the ferrule polishing method, ferrule mechanical contact method, or ferrule fusion contact method. However, the process of arranging the length of the optical fiber, i.e., inserting the optical fiber into the ferrule by hand and checking its cross section, takes 60% of the time required for the entire process, which increases the overall cost. Therefore, in order to make this task more cost-effective, we will develop an automated inspection system with automatic cross-sectional arrangement of a field-installable connector. This system will be able to decrease the failure rate from 10% to 2% compared with the conventional method when cutting the optical fiber inserted into the ferrule. It will also improve the productivity by decreasing the test time by 28% compared with the conventional method. Our studies showed that it was possible to reduce the production costs and improve the quality of a field-installable connector, and we expect it to dominate the market.

• Journal of the Korea Computer Graphics Society
• /
• v.22 no.4
• /
• pp.11-19
• /
• 2016

In this paper, we propose an automatic segmentation method of renal parenchyma on abdominal CT image using graph-cuts with shape constraint based on multi-probabilistic atlas. The proposed method consists of following three steps. First, to use the various shape information of renal parenchyma, multi-probabilistic atlas is generated by cortex-based similarity registration. Second, initial seeds for graph-cuts are extracted by maximum a posteriori (MAP) estimation and renal parenchyma is segmented by graph-cuts with shape constraint. Third, to reduce alignment error of probabilistic atlas and increase segmentation accuracy, registration and segmentation are iteratively performed. To evaluate the performance of proposed method, qualitative and quantitative evaluation are performed. Experimental results show that the proposed method avoids a leakage into neighbor regions with similar intensity of renal parenchyma and shows improved segmentation accuracy.

• The KIPS Transactions:PartB
• /
• v.9B no.4
• /
• pp.445-452
• /
• 2002

In Korean-English mixed documents, translated technical words are usually used with the attached full words or original words enclosed with parenthesis. In this paper, a collective method is presented to recognize and extract the translation phrases with using a base translation dictionary. In order to process the unregistered title words and translation words in the dictionary, a phonetic similarity matching method, a translation partial matching method, and a compound word matching method are newly proposed. The experiment result of each method was measured in F-measure(the alpha is set to 0.4) ; exact matching of dictionary terms as a baseline method showed 23.8%, the hybrid method of translation partial matching and phonetic similarity matching 75.9%, and the compound word matching method including the hybrid method 77.3%, which is 3.25 times better than the baseline method.

• The Journal of the Acoustical Society of Korea
• /
• v.22 no.2
• /
• pp.135-144
• /
• 2003

A method for the automatic segmentation of speech signals is described. The method is dedicated to the construction of a large database for a Text-To-Speech (TTS) synthesis system. The main issue of the work involves the refinement of an initial estimation of phone boundaries which are provided by an alignment, based on a Hidden Market Model(HMM). Multi-layer perceptron (MLP) was used as a phone boundary detector. To increase the performance of segmentation, a technique which individually trains an MLP according to phonetic transition is proposed. The optimum partitioning of the entire phonetic transition space is constructed from the standpoint of minimizing the overall deviation from hand labelling positions. With single speaker stimuli, the experimental results showed that more than 95% of all phone boundaries have a boundary deviation from the reference position smaller than 20 ms, and the refinement of the boundaries reduces the root mean square error by about 25%.

• The Journal of Korean Association of Computer Education
• /
• v.6 no.2
• /
• pp.9-21
• /
• 2003

This paper proposes an automatic construction system for transliteration dictionary from English-Korean parallel corpus. The system works in 3 steps: it extracts all nouns from Korean documents as the first step, filters transliterated foreign word nouns out of them with the language identification method as the second step, and extracts the corresponding English words by using a probabilistic alignment method as the final step. Specially, the fact that there is a corresponding English word in most cases, is utilized to extract the purely transliterated part from a Koreans word phrase, which is usually used in combined forms with Korean endings(Eomi) or particles(Josa). Moreover, the direct phonetic comparison is done to the words in two different alphabet systems without converting them to the same alphabet system. The experiment showed that the performance was influenced by the first and the second preprocessing steps; the most efficient model among manually preprocessed ones showed 85.4% recall, 91.0% precision and the most efficient model among fully automated ones got 68.3% recall, 89.2% precision.