• Journal of Information Processing Systems
• /
• 제18권3호
• /
• pp.411-427
• /
• 2022

Digging historical and cultural information from seals in ancient books is of great significance. However, ancient Chinese seal samples are scarce and carving methods are diverse, and traditional digital image processing methods based on greyscale have difficulty achieving superior segmentation and recognition performance. Recently, some deep learning algorithms have been proposed to address this problem; however, current neural networks are difficult to train owing to the lack of datasets. To solve the afore-mentioned problems, we proposed an SEL-RefineMask which combines selector of feature pyramid network (SEL-FPN) with RefineMask to segment and recognize seals. We designed an SEL-FPN to intelligently select a specific layer which represents different scales in the FPN and reduces the number of anchor frames. We performed experiments on some instance segmentation networks as the baseline method, and the top-1 segmentation result of 64.93% is 5.73% higher than that of humans. The top-1 result of the SEL-RefineMask network reached 67.96% which surpassed the baseline results. After segmentation, a vision transformer was used to recognize the segmentation output, and the accuracy reached 91%. Furthermore, a dataset of seals in ancient Chinese books (SACB) for segmentation and small seal font (SSF) for recognition were established which are publicly available on the website.

• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2022년도 하계학술대회
• /
• pp.265-268
• /
• 2022

Video-based Point Cloud Compression(V-PCC) 부호화기의 세그먼트 정제(Refining segmentation) 과정은 3D 세그먼트를 2D 패치 데이터로 효율적으로 변환하기 위한 V-PCC 부호화기의 핵심 파트이지만, 많은 연산량을 필요로 하는 모듈이다. 때문에 이미 TMC2 에 Fast Grid-based refine segmentation 과정이 구현되어 있으나, 아직도 세그먼트 정제 기술의 연산량은 매우 높은 편이다. 본 논문에서는 현재 TMC2 에 구현되어 있는 Fast Gridbased Refine Segmentation 을 살펴보고, 복셀(Voxel) 타입에 따른 특성에 맞춰 두 가지 조건을 추가하는 고속화 알고리즘을 제안한다. 실험 결과 압축성능(BD-BR)은 TMC2 와 거의 차이를 보이지 않았지만, 모듈 단위 평균 10% 연산량이 절감되는 것을 확인하였다.

• 반도체디스플레이기술학회지
• /
• 제3권3호
• /
• pp.19-21
• /
• 2004

This paper presents a 3-dimensional diffusion simulation with adaptive solution strategy. The developed diffusion simulator VLSIDIF-3 was designed to re-refine areas. Refine scheme was calculated by the difference of doping concentration between any of two nodes. Each element is greater than tolerance and redo diffusion process until error is tolerable. Numerical experiment in low doping diffusion problem showed that this adaptive solution strategy is very efficient in both memory and time, and expected this scheme would be more powerful in complex diffusion model.

• 항공우주기술
• /
• 제8권2호
• /
• pp.75-82
• /
• 2009

본 문서는 일반 사용자들이 다목적실용위성 2호(KOMPSAT-2) 영상을 사용하는데 있어 보다 쉽게 접근할 수 있도록 하기 위해, 기존의 상용 S/W를 이용하여 영상의 센서모델링을 적용하는 방법에 대해 설명하고자 한다. 센서모델링은 다목적실용위성 2호 자료로부터 정량적, 정성적 정보를 추출하기 위한 가장 기본적인 단계라고 할 수 있다. 우선 다목적실용위성 2호(KOMPSAT-2) 영상자료와 함께 제공되는 보조데이터의 종류와 내용에 대해 살펴보고, 이를 상용 S/W에서 어떻게 사용할 것인가에 대해 설명하였다. 센서모델링에 적용은 Polynomial-base의 센서모델링과 Refine RFM을 적용하였다. Polynomial-base 센서모델링에서, 위성위치는 1차 다항식 형태의 등속직선운동, 위성자세는 X축, Y축, Z축에 대하여 고정값 또는 Y축과 Z축에 대한 1차 다항식 형태가 정확도가 우수한 것으로 나타났다. Refine RFM 모델링을 수행할 때 Affine 모델을 적용할 경우 정확도가 1 pixel 이하로 확보됨을 알 수 있었다.

• 한국의류산업학회지
• /
• 제11권5호
• /
• pp.799-807
• /
• 2009

The purpose of this study was to examined which the pursuing image, make-up and hair style desire on their wedding day and how the future brides managed their appearance for the wedding. This qualitative research was based on the data collected by in-depth interviews with 15 brides in their 20s, 30s and 40s who reside in Daegu and Gyeongbuk. The results of the study are shown follow: The images that the brides most pursued for the wedding day were elegant images and refine images. The favorite wedding-dress styles of brides were elegant, refine, lovely, classical, and natural style in order of preference. The design of the dress was top priority when selecting the wedding-dress and it was followed by the function that complements anyones' shape. The brides liked natural style wedding make-up most and refine and young one followed. They considered eye make-up most important after skin expression, that is, covering defects of their skin. Concerning wedding hair styles, an up-style which is typical for wedding is desired most and a natural flowing hair style second. Regarding fitness, they mainly took physical exercise and went on a diet which is not unreasonably harsh.

• 한국전산구조공학회:학술대회논문집
• /
• 한국전산구조공학회 2006년도 정기 학술대회 논문집
• /
• pp.333-340
• /
• 2006

In these days the floor vibration is beginning to make its appearance of the environmental dispute in apartment building. Standard floor system are suggested for the settlement of this issue by government. The sound of floor impact sound is needed to secure comfortable quality in housing. Also, it is required an accurate analysis and a proper evaluation for floor vibration. Refine model is necessary for the floor system of housing to analyze accurately the floor vibration. But this refine model is not efficient because it is required so much running time for vibration analysis and it is difficult of modeling of standard floor slab. In this paper, new modeling methods of standard floor slab are proposed for the accurate rigidity evaluation. By using the new modeling method, the accurate vibration response can be obtained and can accurately evaluate the rigidity of standard floor system with resilient materials. Therefore the proposed modeling method is of practical use for vibration analysis of floor system of apartment building.

• 한국언어정보학회:학술대회논문집
• /
• 한국언어정보학회 2007년도 정기학술대회
• /
• pp.430-439
• /
• 2007

In this paper, we use non-negative matrix factorization (NMF) to refine the document clustering results. NMF is a dimensional reduction method and effective for document clustering, because a term-document matrix is high-dimensional and sparse. The initial matrix of the NMF algorithm is regarded as a clustering result, therefore we can use NMF as a refinement method. First we perform min-max cut (Mcut), which is a powerful spectral clustering method, and then refine the result via NMF. Finally we should obtain an accurate clustering result. However, NMF often fails to improve the given clustering result. To overcome this problem, we use the Mcut object function to stop the iteration of NMF.

• 한국주조공학회지
• /
• 제15권3호
• /
• pp.262-271
• /
• 1995

It is well known what is impossible to refine primary and eutectic Si simultaneously by addition of both Cu-P and Sr(or Na). Because of the formation of compound $Sr_3P_2(or Na_3P)$ in the melt, in the result, both effects disappear. In this study Al-Cu-P alloy that comprises AlP compounds inside is added in the melt with Sr simultaneously. As AlP compounds that act on nucleation sites of primary Si are not formed but added directly an the melt, it is difficult to form $Sr_3P_2$ by reaction with Sr. Therefore it is possible to refine primary and eutectic Si simultaneously in the general casting process by use of Al-Cu-P alloy and Sr.

• 한국건축시공학회:학술대회논문집
• /
• 한국건축시공학회 2015년도 춘계 학술논문 발표대회
• /
• pp.44-45
• /
• 2015

Although the carbonation problem of high-volume slag concrete has been solved by the mixed use of emulsified refine cooling oil as the pilot study, there was an instance of securing air quantity and frost resistance followed by the use of Expancel due to the problem of having vulnerability in frost resistance in result as the air quantity has not been secured. But due to the problem of Expancel not being economical, air quantity and frost resistance of high-volume slag concrete mixed with ERCO are attempted to be secured using micro bubble that can remain in the water for a long time as a relatively economical and very microscopic bubble.

• 천문학회보
• /
• 제44권1호
• /
• pp.83.1-83.1
• /
• 2019

Identifying asteroid families, which are groups of asteroids with similar orbital properties, is important for understanding the formation and evolution of the solar system, and probing the origins of Near-Earth Objects (NEOs). Although asteroid taxonomy can be used to identify and refine asteroid families, there are numerous asteroids which are not taxonomically classified yet. Korea Microlensing Telescope Network (KMTNet) can be useful to investigate types of that asteroids, because the telescope can observe a number of asteroids at once by its large field of view. Using KMTNet data, we confirmed that the taxonomic classification of the asteroids is possible by plotting color-color diagram. There is a clear division between C-type and S-type, but ambiguous division between C-type and X-type. In the future, we will observe and classify asteroids which are not classified yet and utilize the data to identify and refine asteroid families.