• 전기의세계
• /
• v.54 no.6
• /
• pp.24-29
• /
• 2005

• Proceedings of the Korean Institute of Communication Sciences Conference
• /
• 1998.10a
• /
• pp.447-450
• /
• 1998

• Proceedings of the Acoustical Society of Korea Conference
• /
• spring
• /
• pp.65-68
• /
• 2004

본 논문에서는 반복학습으로 음소모델을 강건하게 하여 음소기반 핵심어 검출 시스템의 성능을 개선하고자 하였다. 가변어휘 핵심어 검출 시스템은 인식 대상 핵심어의 추가와 변경이 용이하도록 모노폰 단위로 핵심어 모델과 필러 모델을 구성하였다. 핵심어 모델과 필러 모델은 동일한 음소모델을 이용하므로 각각의 음소 모델의 분별력 향상은 핵심어 검출 성능과 밀접한 관계에 있다. 따라서 본 논문에서는 음소 HMM(Hidden Markov Model)의 학습시에 반복 학습을 통하여 음소 모델을 강건하게 만든 후 핵심어 검출 실험을 수행하였다. 그 결과, 10회의 반복학습을 통하여 얻어진 음소 HMM을 이용한 핵심어 검출의 성능은 반복학습을 하지 않은 경우보다 핵심어 검출의 CA-CR 평균 성능이 $4\%$ 향상됨을 확인할 수 있었다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.11a
• /
• pp.285-289
• /
• 2009

본 논문에서는 현재 비디오 코덱 표준인 H.264/AVC 에서 사용하고 있는 정수형 변환을 두 번 사용하는 다차원 변환 즉, MDT(Multi-Dimensional Transform) 을 제안한다. 제안하는 알고리듬은 H.264/AVC 7가지 인터 모드중에서 $16{\times}16$, $8{\times}16$, $16{\times}8$블록 모드에 적용된다. H.264/AVC에서 사용하는 정수형 변환을 $4{\times}4$블록 단위로 적용하고 인접하는 4개의 $4{\times}4$블록으로부터 같은 주파수 위치의 계수값을 모아서 16개의 $4{\times}1$벡터 변환을 추가로 수행하는 방법이다. MDT를 수행함으로써 H.264/AVC 표준에서 사용되는 변환이 가지는 예측오차를 유지하면서 공간적 중복도를 추가로 줄일 수 있다. MDT 된 계수들을 정수형으로 양자화하고 양자화된 계수값을 효율적으로 주사하는 방법을 제안하여 압축효율을 높였다. 제안된 알고리듬의 제안하는 MDT 기반 방법은 H.264/AVC High profile 표준보다 평균 3.00%의 비트 절감을 보였다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2009.11a
• /
• pp.277-280
• /
• 2009

IPTV, 블루레이 디스크, DMB 등과 같은 멀티미디어 서비스에서 H.264/AVC 비디오 표준기술을 채택하고 있다. 따라서 H.264/AVC 비트스트림을 고속으로 인덱싱하기 위한 썸네일 기술이 요구 된다. 그러나 H.264/AVC는 기존 표준기술과는 다르게 인트라 모드에서도 예측방법을 이용하기 때문에 새로운 썸네일 추출방법이 요구되어 최근에 H.264/AVC 비트스트림 상에서 썸네일을 추출하는 방법이 제안되었다. 그러나 이 방법에서는 인트라 $16{\times}16$ 모드와 연관된 블록에서 심각한 화질의 저하가 발생하며, QP 값이 커질수록 그 증상이 더 심해지는 문제점이 있다. 그리고 공간주파수 영역에서 처리하기 때문에 예측 모드에 따라 연산오류가 발생하여 이 오류가 파급되는 문제가 있다. 따라서 본 논문에서는 공간주파수 영역에서 H.264/AVC 썸네일을 추출할 때 인트라 $16{\times}16$ 예측 모드에서도 오류가 발생하지 않도록 하는 방법과 공간주파수 영역에서 발생하는 연산 오류를 보상하는 방법을 제안한다. 그리고 제안한 방법을 다양한 시험 비디오 시퀀스에 적용하여 이전 썸네일 추출 방법과 비교하여 프레임에 따라 최대 PSNR 약 4dB 증가 및 주관적 화질을 향상시켰다.

• Journal of Broadcast Engineering
• /
• v.14 no.3
• /
• pp.342-356
• /
• 2009

This paper presents a vote decision-based deinterlacing scheme for false directional error correction(VDD) to convert interlaced signal into non-interlaced signal using only one fields. The VDD using the vote decision goes through four steps process. The first step extracts regions having doubt of false edge using MM-ELA method. In these regions, the edge direction is decided by the majority vote using upper adjacent pixels's information through the second step. But, we still have undecided directions, which will be decided by the majority vote and the directional average decision at the third step. This step preserves the edge directions and minimizes visual degradation. Finally, the last step interpolates undecided pixels using DOI method which can consider the fine edge direction. Although the VDD with hierarchical structure has a high complexity, it can extract delicate edge compared to other pixel-by-pixel or window-by-window deinterlacing algorithms. Simulation results show that it has significantly improved both the subjective and objective qualities of the reconstructed images.

• The Journal of the Acoustical Society of Korea
• /
• v.23 no.3
• /
• pp.242-247
• /
• 2004

Conventional keyword spotting systems use the connected word recognition network consisted by keyword models and filler models in keyword spotting. This is why the system can not construct the language models of word appearance effectively for detecting keywords in large vocabulary continuous speech recognition system with large text data. In this paper to solve this problem, we propose a keyword spotting system using pseudo N-gram language model for detecting key-words and investigate the performance of the system upon the changes of the frequencies of appearances of both keywords and filler models. As the results, when the Unigram probability of keywords and filler models were set to 0.2, 0.8, the experimental results showed that CA (Correctly Accept for In-Vocabulary) and CR (Correctly Reject for Out-Of-Vocabulary) were 91.1% and 91.7% respectively, which means that our proposed system can get 14% of improved average CA-CR performance than conventional methods in ERR (Error Reduction Rate).

• (The)Study of the Eastern Classic
• /
• no.70
• /
• pp.257-301
• /
• 2018

The purpose of the study is to examine righteous army activity and combat in Hapcheon during the Imjin invasion period. The resulting conclusions are as follows. First, Chung In-hong raised a righteous army in Hapcheon and Kim Myeon raised an army in Goryeong. Chung In-hong commanded the army and fought against Japanese army with a leadership of knowing the enemy and himself, and Kim Myeon commanded the army and fought against Japanese military with a leadership of harmony. Second, battles of Chung In-hong's righteous army are the first Mugye battle, the ambush battle in Sawon-dong, the Chogye Majin battle, the Aneon battle, and the third Seongju Castle battle. Battles of Kim Myeon's righteous army are the Yeongang battle, the Gaesanpo battle, the second Mugye battle, the Ucheok-hyeon battle, the Jirye battle, and the Sarangam battle. Battles that Chung In-hong and Kim Myeon united and fought are the first and second Seongju Castle battles. In Jeongyu Jaeran, Chung In-hong played a role of Jodosa who takes charge of provisions in the right area of Gyeongsang. In addition, Hapcheon was the access road and transport route where Japanese army entered Jeolla-do. Third, participation and role of Ming troops are part of restraint device against Japanese army with military tactics of ii chei(using foreigners to control foreigners) to remove Japanese army from Joseon and defend Yodong. After a victory of Li Rusong in Pyeongyang Castle in January, 1593(the 26th year of Seongjo), Ming troops pursued practical interest through peace talks rather than active battles. When there was practically Siege of Jinju, Yujeong troops of Ming entered with hand-to-hand martial arts of Sacheon soldiers in late June, 1593, but did not participate because they should wait for orders of Gyeongnyak Song Eung-chang and Admiral Li Rusong. Fourth, in the Imjin invasion period, Joseon suffered terrible damage such as ruin of the whole country by invasion of Japanese army and in need of aid of military rice from Ming troops, but righteous armies and the royal forces in each area cooperated so defeated Japanese army. It is understood through a case of Hapcheon in the right area of Gyeongsang. Especially, Joseon did not succumb to pressure of Ming troops that used full powers to two aspects such as settlement and battle with Japan during the war, and did not lose national confidence and pride by showing a fighting will to fight against Japanese army to the end with independence. Such a spiritual culture originated from homeland protection and loyalty to the king, and is national spirit of resistance that sublimates the united mind and spirit of community to protect a country against foreign invasion.

• Korean Language and Literature
• /
• v.67 no.67
• /
• pp.45-71
• /
• 2018

Songgubong is a person in the mid-Choseon Dynasty, but he is a person who is exalted as a teacher of four generations. However, the life of Songbong was a series of trials of 'returning' due to the environment and political changes born in prestige. There was also a political upheaval in the era where he lived, such as a Sinsamuok, Kichukoksa the beginning of Parts. Songgubong also shared his friendship with YulGon, Sounhorn in his life. This paper examines the narrative style of the Songgwon legend, who was a real person of the Choseon Dynasty, and examines the historical meaning of the Gubong legend. In this paper, 40 pieces of Songbong were analyzed. So I made 20 pixels and 5 narrative types. The basic form of the Songgwangbong legend is 'the exchange between the man and the gongbong - the complaints of the surrounding men - Songbongbang face to face and overwhelming'. This basic type is expanded to an anger that strengthens Songbong's greatness, an anecdote that worsens Songbong's desire, and an anger that emphasizes the misery of Songbongbong. Following the basic type and the extended type, the comprehensive type is combined with the active play. Actually, Songbong was in exile and could not participate in the war. However, the legendary Songbong is short enough to shorten the duration of the imprisonment, or to give the Turtle Ship to Yi Soon Shin. The story of the ending of the Imran in a few months if it has been used properly has been handed down on the Internet. Song Hyung-bong, who is a politically talented person, plays an active role as a novi in the story. Songbong is a "hero" who escapes from the constraint of "novi" and interacts with them in a similar way, controlling them with "eyes" and charisma. Song Jung - bong 's performance succumbs to Lee Yong Song and controls Yulgok and Yi Soon Shin. In addition, the legend of Jinan, which can be said to be the scene of Ogasaki, is still handed down to the dignity of Songbong. The evaluation of the Songbong who made a stroke of the company is interpreted as a political sacrifice in narrative. Compared to the spiritual virtue of the person who interacted with him, the marriage, and Jungchol, Songbong was the one who was able to withstand the years of vigor due to the occasional ox. The meaning of the Songgwon legend in the Conflict is that, apart from the partisan lessons that had to be taken from the existing records, the people are worshiping him as the epitome of the great weak which is excluded from real power. I hope that this legend will make it possible to look at the existing evaluation from a new angle.

• Journal of Broadcast Engineering
• /
• v.13 no.5
• /
• pp.622-635
• /
• 2008

Recently, high definition media services like HDTV and IPTV are growing. A fast reduced-size image extracting method is required to meet what those services require. Conventional DC image extracting methods, however, can't be applied to H.264/AVC streams since a spatial domain prediction scheme is adopted in H.264/AVC intra mode. To solve this problem, a thumbnail extraction method in H.264/AVC was proposed. However, the method has mismatch problem which was caused by round-off operation in intra prediction and mismatch between integer and floating point calculation. In this paper, we propose an error compensation method for extracting thumbnail directly in H.264/AVC bitstreams. The compensation method introduces the mismatch problem in thumbnail extraction and presents compensation values. Through the implementation and performance evaluation, proposed method compensated round-off error efficiently in D1 and HD sequences while the additional extraction time is negligible.