본 논문은 키넥트 센서에서 랜덤 무향 칼만 필터에 기반한 잡음 제거를 통하여 신체 골격을 추정하는 방법을 제안한다. 일반적인 RGB 값과 깊이 정보를 제공하는 키넥트 카메라는 시간 영역에서 변동하는 센서 응답으로 인해 잡음이 포함된다. 기존의 방법은 다양한 필터링 기법을 사용하여 잡음 제거를 시도하였으나 잡음의 비선형성 때문에 한계가 있었다. 따라서 본 논문에서는 비선형 잡음 특성을 예측하고 업데이트하기 위하여 랜덤 무향 칼만 필터를 적용하고 이를 바탕으로 자세 인식을 위한 3 차원 공간에서 신체 관절 포인트를 예측하였다. 실험 결과 제안한 방법은 기존의 방법보다 잡음 감소 및 뼈대 추정에서 우수한 성능을 가지는 것을 확인하였다.

The purpose of this study was to create canine skeleton modeling files (obj., STL., etc.)for more accurate canine kinesiology and physical rehabilitation. Firstly, 3-dimensiona (3D) scanning on real naturally passed away dog bones and handling with additional reverse engineering process were conducted. This study indicated that the final modeling files could contribute to improve initial movements of canine rehabilitation or academic filed of kinesiology in Korea.

The main focus of this study was to create canine skeleton modeling files and real-world structures for canine kinesiology and physical rehabilitation education. Firstly, 3-dimensional (3D) modeling was conducted with additional reverse engineering process, and then directly fabricated the files. In other way, we made multi-joint based canine structure using conventional corrugate tube. This study indicated that the final real-world structures could contribute to improve initial movements of canine rehabilitation or academic filed of kinesiology in Korea.

엽연의 모양이 다른 4종류의 잎을 식물재료로 사용 하고, 망사잎 제작을 위한 엽육제거시 sodium hydroxide( NaOH)의 농도와 처리시간의 최적 조건을 찾고자 하였다. 중국단풍(장상열)은 NaOH 60% 용액에서 50 분간, 호랑가시나무(예거치)는 NaOH 20% 용액에서 50분간, 신갈나무(천열)는 NaOH 20-60% 용액에서 70분간, 대왕참나무(전열)는 NaOH 20-40% 용액에서 70분간 열탕 처리하다 엽육제거율도 높고 관상가치가 높은 망사잎을 만들 수 있었다.

In spite of relatively high prices, aluminum wheels rapidly replace traditional steel wheels because of light weight and stylish design. To cope with expanding aluminum wheel market, steel wheel makers try to produce skeleton typed steel wheels adopting high strength steel which are light, stylish and lowprice compared with conventional steel wheel. But skeleton typed steel wheel disks are difficult to form because of complex shapes and low drawability of high strength steel. This paper presents the drawing process design for 5-spoke skeleton steel wheel disk. Through the finite element simulation of various alternatives for multi stage drawing processes, desirable draw die designs were proposed.

The present work has attempted to investigate the dependence of Cu infiltration kinetics on in-situ structure modification of W powder skeleton in W-Cu system. In-situ structure modification of W skeleton by addition of 0.3wt%Ni-P eutectic alloy was designed to proceed during heat-up of the W compact for Cu infiltration process. It was found that the Ni-P added W skeleton underwent remarkable stucture change only during heating-up. its structure was composed of large necks of W particles above 0.5 in the ratio of neck to particle size and smooth pore channels. The infiltration experiment showed that the infiltration kinetics for the W-Ni-P followed well the linear relationship of h vs. the rate constant K of which was in good agreement with the theoretical value. On the other hand, in case of the pure W skeleton a lower K value by 20% than the theoretical one was obatined. Such discrepancy is discussed in terms of skeleton structure induced infiltration mechanics.

본 논문에서는 스켈레톤(skeleton)의 리깅(rigging)방식을 사용하여 해부학을 기초로 한 14개의 표정근과 동일한 기능을 수행하게 얼굴의 표정을 표현하는 SSFE(Skeleton System for Facial Expression)를 제안한다. SSFE는 3차원애니메이션저작도구(MAYA 8.5)에서 제공하는 스켈레톤을 Deformer로 이용하여 메시모델(mesh model)의 눈, 코, 입 주변의 얼굴표정을 단위 동작별로 구분하여 조합하여 얼굴표정을 만들어 내는 방식이다. 이 결과, SSFE식 스켈레톤은 다른 모델에 무한대로 변형하여 재활용 할 수 있게 해준다. 이러한 재활용성은 애니메이션의 제작방식의 OSMU(One Source Multi Use)로 이해 될 수 있으며 애니메이션의 제작비용을 줄일 수 있는 대체방식이다. 또한 3차원 애니메이션 관련업체 즉 가상현실이나 게임, 제작에 활용될 수 있다.

Mathematical morphology skeleton image processing makes many partial skeleton image planes from an original binary image. And the original binary image can be reconstructed without any distortion by summing the first partial skeleton image plane and each dilated partial skeleton image planes using the same structuring element. Especially compression effects of Elias coding to the morphological globally minimal skeleton(GMS) image, is better than that of PCX and Huffman coding. And then this paper proposes mathematical morphological GMS image processing which can be applied to a binary image transmitting for facimile and big size(bigger than 64×64 size) bitmap fonts storing in a memory.