• 한국시뮬레이션학회논문지
• /
• 제10권3호
• /
• pp.95-103
• /
• 2001

Rendering of 3D terrain data in real-time is difficult because of its large scale. So, it is necessary to use level-of-detail(LOD) that uses fewer data, but makes almost similar image to the original. We present an algorithm for real-time LOD generation and rendering of 3D terrain data. The algorithm applies wavelet transform to the terrain data, and then generates quadtree based view-dependent LOD using wavelet coefficients that are the output of wavelet transform. It also uses frame-to-frame coherence and view culling for high frame rates.

• 한국게임학회 논문지
• /
• 제8권3호
• /
• pp.61-68
• /
• 2008

사진트리 기반의 지형 시각화 기법은 많은 응용 프로그램에서 활용되어 왔다. 하지만 전체 과정이 CPU에서 수행되기 때문에 GPU를 사용하는 다른 방법들에 비해 렌더링 성능이 떨어진다. 본 논문에서는 사진트리 기반의 지형 시각화 기법을 GPU에서 수행할 수 있도록 오차텍스처와 LOD텍스처를 제안하고, 상세단계가 적용된 사진트리 블록을 동일한 해상도의 메쉬로 채워서 렌더링 속도를 향상시키는 방법을 제안한다. 전처리 단계에서는 보편 공간에서 사진트리의 연속된 두 단계사이에서 지형의 높이 값 차이를 계산하여 오차텍스처에 저장한다. 렌더링 단계에서는 저장된 오차 값을 이용하여 투영된 오차 값을 계산하고, 그 결과를 LOD텍스처에 저장한다. LOD텍스처에 저장된 값을 이용해서 블록단위로 시각 절두체 선별을 하고 상세단계를 선택한다. 이 방법은 부하가 큰 상세 단계 선택 작업을 GPU에서 수행하고 블록단위 연산을 함으로써 작업량을 줄일 수 있다. 상세 단계가 서로 다른 블록이 인접해 있을 경우 T-정점 때문에 크랙이 발생하는데 원본 고도 데이터의 밉맵을 활용해서 이것을 제거할 수 있다.

• 한국산학기술학회논문지
• /
• 제16권10호
• /
• pp.6917-6924
• /
• 2015

인간의 시각구조는 시야각과 가시거리에 의해 형성된다. 그러나 시지각에 기반한 대다수의 공간구조분석이론은 시야각을 반영하고 있지 않거나 평면적인 시야각만을 고려하고 있다. 일부 반영하는 이론들도 시야각에 포함된 부분과 제외된 부분을 이분법적으로 분리시킨 이론으로서 인간의 보편적 주시특성이 배제되어 있다. 본 연구는 아이트래킹 실험을 통하여 시야각의 한계를 실증적으로 파악하고 시야각별 주시확률을 도출하여, 이를 3차원 가시성 분석이론에 적용하였다. 그리고 시야각체 제한이 가능한 공간 시지각 분석어플리케이션을 제작하고 여기에 시야각별 Multiple frustum culling의 개념을 적용하여 그 가능성을 파악하였다. 주시특성 파악은 33명을 대상으로 이동형 아이트래커를 이용하여 임의의 공간에 대한 3분간의 보행과정을 측정한 데이터를 수집하였으며, 각도별로 주시고정 빈도 확률을 분석하여 데이터를 마련하였다.

• 한국임상수의학회지
• /
• 제33권2호
• /
• pp.93-96
• /
• 2016

This retrospective study evaluated the effect of age at first calving on the productive and reproductive performance in Holstein dairy cattle. Data were collected on 2,297 lactations from 19 dairy farms, including dates of birth, insemination, pregnancy, calving, and lactation and milk production (305-day milk yield at first lactation and lifetime milk yield). Lactation data were grouped based on age at first calving into the early (< 24 months, n = 414), moderate (24 ${\leq}$ and ${\leq}$ 28 months, n = 1,268), and late (> 28 months, n = 615) groups. The 305-day milk yield at first lactation was higher (p < 0.005) in the late group (8,461.9 kg) than in the early (8,130.8 kg) and moderate (8,261.9 kg) groups. Lifetime days in milk were shorter (p < 0.01) in the early group (1,045.7) than in the moderate (1,143.1) and late (1,110.7) groups, whereas the lifetime milk yield was higher (p < 0.05) in the moderate group (34,171.8 kg) than in the early group (31,791.6 kg). The second calving interval (days) differed (p < 0.0001) between the early (416.9), moderate (438.9), and late (455.3) groups. The culling rate at the first and second lactations tended to be higher (p = 0.08) in the early group (31.2%) than in the moderate group (26.0%), whereas it was intermediate in the late group (29.3%). In conclusion, dairy heifers aged 24-28 months at first calving showed a higher productive performance through an increased lifetime milk yield and a decreased culling rate.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 1999년도 가을 학술발표논문집 Vol.26 No.2 (2)
• /
• pp.595-597
• /
• 1999

기존의 LOD 제어방법들은 랜더링속도를 성공적으로 증가시켜왔으나 오버헤드가 크다는 단점을 갖고 있다. 이러한 오버헤드는 각 vertex마다 view-frustum clipping, back-face culling, 스크린 공간 기하학적 오차계산과 같은 view-dependent refinement criteria를 측정하고, 메쉬의 LOD를 바꾸기 위해서 edge collapse/vertex split를 수행하기 때문이다. 제안하는 방법은 메쉬를 여러 개의 region들로 나누고 vertex가 아닌 region에 대해 view-dependent refinement criteria를 측정하므로 오버헤드가 훨씬 작다. 또한 각 region 들의 LOD가 바뀔 때 미리 만들어 둔 LOD 버전들중에서 하나를 선택하기만 하면 되므로, edge collapse/vertex split을 수행하는 오버헤드는 없다. 실험적으로 제안하는 LOD 제어방법은 기존의 방법들보다 작은 메모리를 사용하고 LOD 제어 오버헤드도 적으며, LOD 제어를 하지 않은 경우보다 2배-5배의 랜더링 속도향상을 얻었다.

• 한국동물위생학회지
• /
• 제35권4호
• /
• pp.255-261
• /
• 2012

Foot and mouth disease (FMD) is highly infectious disease of cloven-hoofed animals, particularly cattle, sheep, pigs and goats. Last outbreak reported in November, 2010 induced the enormous social and economical impacts. Culling of infected animals, movement control, and vaccination are the major control measures of FMD. The aim of this study was to detection foot-and-mouth disease virus (FMDV) in the soil and leachate from modeling burial for pig carcass as measured by real-time reverse transcriptase polymerase chain reaction (RT-PCR). FMDV and Coxsakievirus B1 (CVB1) were detected in soil by week 16 and Coxsakievirus B1 (CVB1) by weeks 12, respectively. FMDV and CVB1 also detected by weeks 8 in the leachate. Results from this study provides an evidence that FMDV could be inactivated for safe of pig carcasses infected with FMDV within 4 month in the carcass burial site.

• 농촌지도와개발
• /
• 제16권4호
• /
• pp.939-965
• /
• 2009

The Machine Learning has been identified as a promising approach to knowledge-based system development. This study aims to examine the ability of machine learning techniques for farmer's decision making and to develop the reference model for using pig farm data. We compared five machine learning techniques: logistic regression, decision tree, artificial neural network, k-nearest neighbor, and ensemble. All models are well performed to predict the sow's productivity in all parity, showing over 87.6% predictability. The model predictability of total litter size are highest at 91.3% in third parity and decreasing as parity increases. The ensemble is well performed to predict the sow's productivity. The neural network and logistic regression is excellent classifier for all parity. The decision tree and the k-nearest neighbor was not good classifier for all parity. Performance of models varies over models used, showing up to 104% difference in lift values. Artificial Neural network and ensemble models have resulted in highest lift values implying best performance among models.

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2004년도 춘계학술대회 논문집
• /
• pp.93-98
• /
• 2004

A reasonable analysis of cutting force in end milling may give much advantage to improvement of productivity and cutting tool life. In order to analyze cutting force, the cutting dynamics was modelled mathematically by using chip load, cutting geometry, and the relationship between cutting forces and the chip load. Specific cutting constants of the cutting dynamics model were obtained by average cutting forces, tool diameter, cutting speed, feed, axial depth, and radial depth of cut. The effects of the cutting conditions on the specific cutting force constants in milling were studied. The model is verified through comparisons of model predicted cutting forces with measured culling forces obtained from machining experiments

• 한국공작기계학회:학술대회논문집
• /
• 한국공작기계학회 2004년도 춘계학술대회 논문집
• /
• pp.88-92
• /
• 2004

We developed a helical type drum with an oblique cutting condition for sawdust machine. The existing drum has an orthogonal cutting(2D) condition considering of saw cutter and wood cutting pattern. Additional disadvantage of this type include big cutting resistance, high cost of cutting power and extreme occurrence of vibrations. To improve this shortcoming, we developed a helical type drum with an oblique cutting(3D) condition, therefore, this type drum decreased the vibration and cutting resistance of sawdust machine, and also improved productivity and sawdust porosity.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2002년도 춘계학술대회 논문집
• /
• pp.405-408
• /
• 2002

A high speed feeding type laser cutting UC is developed for the precise cutting of sheet metal. The M/C is a flying optics type one with specified functions of movability and strength in order to increase workability. The gantry frame should be moved with a certain velocity within a relatively short time for the proper cutting of object materials. The gantry is actuated by rack and pinion. In this paper, modal and structural analysis for a laser culling M/C, is carried out in development of the machine The machine is modeled by placing proper shell and solid finite elements and fictitious mass properties to represent the real one.