(A) study on biologically inspired model for object recognition

루옌펑 Korea University 2015 국내박사

In this dissertation, we enhance a biologically inspired hierarchical model for object recognition, Hierarchical Model and X (HMAX), by variety of proposed methods. We design an Oriented Gaussian-Hermite Moment (OGHM) and an OGHM-based HMAX (OGHM-HMAX). The OGHM is an extension of the modified discrete Gaussian-Hermite moment (MDGHM) with an oriented movable mask that allows directional features of images to be represented. The directional features follow the mechanism in the primary visual cortex that responds strongly to oriented bars and has strong phase dependence. We also propose an application of the accumulated OGHM to obtain distinguishable feature information in the first stage of HMAX (S1 layer). As a result, the proposed method can present more selective and robust features and that outperforms the conventional HMAX model by significantly improving the classification accuracy. In addition, the conventional HMAX model often brings in redundant extraction and high computational cost caused by random-patch-selection. To address this issue, we propose a novel patch selection method for HMAX called Saliency and Keypoint based Patch Selection (SKPS), which is based on a saliency (attention) mechanism and multi-scale keypoints. In contrast to the conventional random-patch-selection based HMAX model that involves huge amounts of redundant information in feature extraction, the SKPS based HMAX model (S-HMAX) extracts a very few features while offering promising distinctiveness. To show the effectiveness of S-HMAX, we apply it to object categorization. The experimental results demonstrate that S-HMAX outperforms conventional HMAX and is very comparable to existing architectures that have a similar framework. Although HMAX in object recognition is robust, scale and shift invariant, it has been shown to be sensitive to rotational deformation. To address this, we propose a novel Dominant Orientation Patch Matching (DOPM) that calculates the dominant orientation of selected patches and achieve a patch-to-patch matching. We also propose an application of the DOPM in the HMAX model (D-HMAX) to obtain a rotation invariant feature representation in the third stage of HMAX (S2 layer). As a result, the D-HMAX achieves better robustness in the rotation deformation of input images and that outperforms the conventional HMAX model by significantly improving the classification accuracy.

Designing a novel visual tracking framework for a vehicle vision system

최인환 Korea University 2016 국내박사

This dissertation proposes step-by-step processes to build a visual tracking framework and applies the framework to a vehicle vision system. Visual tracking is still a crucial and challenging task in many computer vision applications and is related to various processes. In this dissertation, a novel visual tracking framework composed of moment-based feature extraction techniques, detection with feature fusion, and tracking algorithms is introduced and applied to a vehicle vision system. The features contain significant information related to solving a computational task for a certain goal. Therefore, the task can be properly carried out by utilizing these reduced representatives, which are called feature descriptors. In this dissertation, a moment-based speed up robust features (SURF) descriptor called the modified discrete Gaussian-Hermite moment SURF (MDGHM-SURF) and a moment-based histogram of oriented gradients (HOG) descriptor called the simplified discrete Gaussian-Hermite moment (SDGHM-HOG) are proposed and extensive experiments are carried out to demonstrate their advantages. Another important process of visual tracking is object detection. This is computer vision technology that deals with detecting the existence of wanted objects of a certain class, such as humans, signs, and vehicles in digital images. Model-based object detection using image features is widely used on account of its advantages, such as easy learning and general performance. However, it is insufficient for use in complex environments. To overcome this incompleteness, this dissertation proposes an object detection framework including a new feature fusion method based on the adaptive neuro-fuzzy inference system (ANFIS) to compensate for the respective object model built with different features. Visual object tracking is used to monitor an object undergoing spatial and temporal changes during a video sequence. The Kalman filter (KF) has been widely used on account of its simple structure and easy implementation. Nevertheless, tracking algorithms based on the KF have flaws degrading their performance caused by model parameter uncertainties and sudden motion changes of the target object. To overcome these problems, this dissertation proposes a finite memory tracker (FMT) based on the optimal unbiased finite memory filter (OUFMF). Finally, to demonstrate and emphasize the performance of the proposed step-by-step algorithms, the proposed tracking framework is applied to a vehicle vision system. To provide comprehensive evaluation results, this dissertation adopts several experimental conditions, including harsh driving environments, for each step. In the detection step, a feature fusion algorithm based on the ANFIS is applied to compensate for each feature-based vehicle model. After that, in the tracking step, the FMF using the OUFMF is used to predict target vehicle positions to overcome abrupt target movement. The experiment results prove that the proposed framework provides accurate vehicle tracking results.

Dense flow field algorithm using A binary descriptor and A modified energy function

오현찬 Korea University 2015 국내석사

In this thesis, we describe the development of a dense flow field algorithm using a binary descriptor and a modified energy function. In the dense SIFT flow, the dense SIFT descriptor and an energy function are employed to make the flow vectors containing the movement information of each pixel. The matching process uses descriptor information and a message-passing method in a coarse-to-fine scheme. The matching performance of dense SIFT flow is good for detecting the movement of each pixel. However, large computational time is needed. To reduce the complexity of the description part, the proposed method employs a binary descriptor. The process of the binary descriptor is simple enough to reduce the complexity. However, a conventional energy function is designed for the integer descriptor vector. The energy function must be modified for the binary descriptor. Additionally, the unfair displacement term of the conventional energy function is replaced with a fair displacement term. The evaluation and application demonstrate that the proposed method is faster than the conventional method with respect to making flow field and more robust with respect to diagonal movements.

New state estimation algorithms based on FIR filters for state-space models

박정민 Korea University 2015 국내박사

This dissertation proposes new state estimation algorithms based on finite impulse response (FIR) filters for state-space models. Five different state estimation algorithms, three of which are associated with adaptive techniques for the horizon size in FIR filtering and two of which are associated with hybrid filtering, combination of nonlinear FIR filters and particle filters (PFs), are proposed in this dissertation. One of the most important current issues in FIR filtering is to manage the horizon size because it is a significant parameter that influences the FIR filter's estimation performance. Since the FIR filter's noise suppression ability, tracking speed, and computational complexity are dependent on the horizon size, selecting a suitable horizon size has been one of the most important and challenging problems in FIR filtering. This dissertation proposes three algorithms that automatically adjusts (adapts) the horizon size in order to improve estimation performance of FIR filters. The three proposed algorithms are called the horizon group shift (HGS) FIR filter, weighted average extended FIR filter bank (WAEFFB), and switching extensible FIR filter bank (SEFFB), respectively. Numerical examples verify that the proposed FIR filters using adaptive horizon sizes exhibit smaller estimation errors than standard FIR filters using a fixed horizon size. In addition, better performance of the HGS-FIR filter is demonstrated in comparisons with the existing nonlinear filters including the extended Kalman filter (EKF), unscented Kalman filter (UKF), and PFs. Another important issue in FIR filtering is to combine FIR filters with infinite impulse response (IIR) filters. State estimators with IIR structure, such as Kalman filters (KFs) and PFs, use all past input/output information to produce state estimates. In contrast, FIR filters use only recent finite input/output information. Since FIR filters and IIR filters have complementary characteristics, combining these two types of filters results in a new type of filter that have advantages of the both filters. This dissertation proposes novel hybrid filters, called hybrid particle/FIR filters. In the structure of our hybrid filter, a PF plays a role of main filter that produces state estimates in normal situations. When the PF failures occur owing to the sample impoverishment phenomenon, an FIR filter recovers the PF from failures. This dissertation proposes two types of hybrid particle/FIR filters. One is the combination of regularized particle filter (RPF) with the extended minimum variance FIR (EMVF) filter. The other one is the combination of RPF with the extended unbiased FIR (EFIR) filter. The resulting filters are called the hybrid RP/EMVF filter and the hybrid RP/EFIR filter, respectively. To demonstrate the performance of the proposed state estimation algorithms, several applications are presented. First, the SEFFB is applied to the F-404 engine system and the ground moving target indicator (GMTI) tracking system to demonstrate its performance. Second, indoor human localization based on wireless sensor network (WSN) demonstrates performances of the HGS-FIR filter and the hybrid RP/EMVF filter. Lastly, indoor mobile robot localization based on WSN demonstrates performances of the HGS-FIR filter and the hybrid RP/EFIR filter.

학교장의 서번트 리더십이 교사의 직무만족을 매개로 효과적인 수업행동에 미치는 영향

송묘용 국민대학교 대학원 2013 국내박사

Abstract The Structural Effect of the Principal's Servant Leadership on the Teacher's Effective Teaching Behavior through the Teacher's Job Satisfaction by Song, Myo-yong Major in Educational Administration Department of Education Graduate School, Kookmin University A. The Purpose of the Study The purpose of this study was to clarify the structural effects of the principal’s servant leadership on the teacher’s effective teaching behavior through the teacher’s job satisfaction. The following research subjects were set up in order to accomplish the purpose. First, are there differences in the principal’s servant leadership, the teacher’s job satisfaction or the teacher’s effective teaching behavior depending on the teacher variables(gender, status, educational and career level) and the school variable(the size of schools)? Second, how are the principal’s servant leadership, the teacher’s job satisfaction and the teacher’s effective teaching behavior correlated? Third, is the structural equation model of the principal’s servant leadership, the teacher’s job satisfaction and the teacher’s effective teaching behavior appropriate? Fourth, what is the connection of the structural effects of the principal’s servant leadership upon the teacher’s effective teaching behavior through the teacher’s job satisfaction? B. The Methods of the Study In this study, the principal’s servant leadership was established as the independent variable, the teacher’s job satisfaction was established as the mediator variable, and the teacher’s effective teaching behavior was established as the dependent variable. The measurement tools were surveys of the principal’s servant leadership, surveys of the teacher’s job satisfaction, surveys of teacher’s effective teaching behavior, with a five-point Likert scale used for the questions of each survey. In order to collect the data for this study, 1,095 sets of surveys were distributed to the teachers of 33 elementary schools in 11 District Offices of Education in Seoul and 967 sets were returned. Except 44 unreliable sets of survey which were not completed or collectively slanted, 923 sets of surveys were used for the data analysis. The data collected through the surveys were analyzed and verified through t-test, one-way ANOVA, Pearson’s correlation coefficient, Structural Equation Model by SPSS Win 18.0 and AMOS 18.0 programs. C. The Results of the Study The essential results of the analysis are as follow. First, the perception level of elementary school teachers in Seoul of the principal’s servant leadership was presented as high in general(M=4.200). More senior teachers than regular teachers, more teachers with longer careers, teachers who work at small sized schools rather than teachers who work at medium or large schools were rated higher in the perception level of the principal’s servant leadership(p<.001). Second, elementary school teachers in Seoul were analyzed as high in the perception level of the teacher’s job satisfaction(M=3.858). More senior teachers than regular teachers, more teachers with longer careers were presented as higher in the teacher’s job satisfaction(p<.001). Third, elementary school teachers in Seoul were graded high in the perception level of their effective teaching behavior(M=4.098). Female teachers and senior teachers were presented higher in the perception level of their effective teaching behavior than male teachers or teachers with shorter careers (p<.05 or p<.001). Fourth, in the analysis, the principal’s servant leadership, the teacher’s job satisfaction, and the teacher’s effective teaching behavior were seen as being correlated with each other. Therefore, it was determined that the establishment of each potential variable and observation variable, and the format of the surveys were appropriate (p<.05 or p<.01). Fifth, the Structural Equation Model(SEM) of the principal’s servant leadership, teacher’s job satisfaction, and teacher’s effective teaching behavior was easily accepted because it was presented that the SEM satisfied almost the entire suitability index. Sixth, the principal’s servant leadership has a direct influence on teacher’s effective teaching behavior(path coefficient=.192), and an indirect influence through teacher’s job satisfaction(path coefficient=.280). Therefore, total influence of the principal’s servant leadership on the teacher’s effective teaching behavior is presented as .472(p<.001). Seventh, the principal’s servant leadership has a direct influence on the teacher’s job satisfaction (path coefficient=.457), the teacher’s job satisfaction also has a direct influence on the teacher’s effective teaching behavior (path coefficient=.613). In addition, it was presented that the teacher’s job satisfaction has a statistically similar intermediary influence on the principal’s servant leadership and the teacher’s effective teaching behavior(path coefficient=.280, p<.001). D. Conclusion The conclusions about the analysis of the structural effects of the principal’s servant leadership on the teacher’s effective teaching behavior through the teacher’s job satisfaction are as follows. First, the principals of elementary schools in Seoul demonstrate high level of servant leadership in general. Second, teachers of elementary schools in Seoul are generally satisfied with their teaching life and effective teaching behavior and they are also trying to improve the teaching quality. Third, the principal’s servant leadership, teacher’s job satisfaction, teacher’s effective teaching behavior are all interrelated affirmatively. Fourth, the principal’s servant leadership has a direct influence on teacher’s effective teaching behavior, and an indirect influence through teacher’s job satisfaction. Fifth, the principal’s servant leadership has a direct influence on the teacher’s job satisfacti

한국과 미얀마의 동물 속담 비교 연구

MYO MYO SWE OO 연세대학교 일반대학원 2017 국내석사

As the exchange between Korea and Myanmar becomes more and more active in the fields of politics, economy, society, culture, tourism, and academia, the number of academic researches on the relationship between the two countries is increasing. However, research reports and tour guidebooks are mostly made from an economic point of view. The number of papers studied from the viewpoint of linguistics among the academic research papers is relatively small, and the research that has been conducted is mostly grammatical research. There is also a comparison between Korean and Myanmar in terms of customs, religion, and language, but there is only one comparative study of proverbs. Therefore, this study aims to compare the proverb of Korea and Myanmar. Compared to the proverbs of both countries, comparing animals by subject, we compared how many animals appeared in both countries. After that, the top 10 animals among the animals of both proverbs were selected, and five common animals were selected and their symbolic meanings were compared. The purpose of this study is to summarize the similarities and differences between the two animal proverbs. This paper is organized as follows. In Chapter 1, the necessity and purpose of research, research subjects and research methods are presented and previous studies are reviewed. In Chapter 2, we discussed basic definitions, features, and functions of proverbs, as well as methods of generating proverbs. In Korea, there are proverbs, slang, and Ian, which means 'deceived words'. In Myanmar, the proverb is zagaboun. Myanmar's proverb name is different from Korea, but the definition of the proverb is similar to that of Korea. In Chapter 3, animal proverbs among the proverbs of the two countries were extracted and classified by materials. We looked at the characteristics of the animals that appeared the most in both countries and the causes that appeared many. The Korean proverbial material is the proverbial dictionary of the Lee Kun Moon (1980), and the proverbial material of Myanmar is Hla Thaung (2014). In Chapter 4, the top ten animal proverbs of the two countries were selected to compare the symbolic meanings of the five common animals. In conclusion, we tried to clarify the limitation of this study while reaffirming the purpose of the study. Since the animal proverb in thispaper is limited to literary considerations, how the speaker's intentions in relation to speech act in the actual speech scene is realized remains a challenge in the future. Nonetheless, the significance of this paper is that it has been able to understand the characteristics of language, lifestyle, and culture between the two countries through identifying common symbolic meanings and differences in animal proverbs. 한국과 미얀마의 교류가 정치, 경제, 사회, 문화, 관광, 학문 분야 등에서 점점 활발해져 가면서 양국 간의 관계에 대한 학술 연구의 수도 늘어나고 있는 추세이다. 그러나 학술 연구논문 중 언어학의 관점에서 연구된 논문의 수는 상대적으로 적으며, 언어학 분야에서 이루어진 연구도 주로 문법 관련 연구이다. 또한 한국과 미얀마 간에 풍속, 종교, 언어 등을 비교한 논문은 있으나 속담을 비교한 연구는 하나 밖에 없다. 따라서 본 연구는 한국과 미얀마의 속담을 비교하고자 한다. 양국의 속담 중 동물을 대상으로 소재별로 비교하였다. 그 후, 양국 속담의 전체 동물 중에서 상위 10가지 동물들을 뽑아 공통적으로 나타난 동물 5가지를 골라 그 상징의미를 비교하였다. 이렇게 함으로써 양국 동물관의 공통점과 차이점을 정리하는 것을 연구의 목적으로 하여 본 논문을 다음과 같이 구성하였다. 1장에서는 연구의 필요성과 목적, 연구 대상과 연구 방법을 제시하고 선행연구를 검토하였다. 2장에서는 속담의 기본적인 정의와 특징, 기능과 더불어 속담의 생성 방법을 살펴보았다. 한국에서 속담을 뜻하는 단어로는 속담, 속어, 이언 등이 있고 이는‘속된 말’이라는 뜻이다. 미얀마에서 속담을 가리키는 말은 zagaboun 이다. 미얀마의 속담 명칭은 한국과 다름에도 불구하고 속담에 대한 정의는 한국과 비슷하게, ‘조상에서 내려오는 교훈 말’이다. 3장에서는 양국의 속담 중 동물 속담을 뽑아 소재별로 분류하여 비교해 보았다. 양국에서 가장 많이 등장한 동물의 특성과 많이 등장한 원인을 살펴았다. 한국의 속담 자료는 이기문(1980)의『속담사전』을, 미얀마의 속담 자료는 Hla Thaung(2014)의 『미얀마속담종합사전』을 대상으로 하였다. 4장에서는 양국 전체 동물 속담의 상위 10가지를 뽑아 공통적으로 나타난 5가지 동물의 상징의미를 비교하였다. 결론에서는 연구의 목적을 재확인하면서 본 연구의 한계점도 밝히고자 하였다. 본고에서 살펴본 동물 속담이 문헌적인 고찰에만 그쳤기 때문에, 실제 발화 현장에서 발화 행위와 관련된 양국 문화 안에 담겨있는 화자의 의도가 어떻게 구현되는가에 대한 깊이 있는 탐색은 앞으로의 과제로 남는다. 그럼에도 불구하고 동물 속담의 공통적인 상징의미와 차이를 확인하는 작업을 통해 양국 간의 언어, 생활 풍습, 문화의 특징을 이해할 수 있었다는 데에 본 논문의 의의가 있다.

Empirical Test for Rice Price Behavior in Asian Countries : A Study of Myanmar, India, Thailand and Vietnam

Myo, Aye Thiri Kangwon National University 2019 국내석사

가격은 농업 생산에 주요 인센티브로 작용하고 있으며, 마케팅 프로세스에 관여하는 모든 참가자의 의사 결정 과정에 대한 시그널이다. 농산물 가격은 생산, 소비자의 복지 및 국가의 수출입을 결정하는 기본적인 정보이기도 하다. 따라서 가격반응에 대한 연구는 시장의 움직임을 이해하는 데 효과적이다. 본 연구는 일반적인 시계열 분석과정 (고정 검정, 공적분 검정, Granger 인과 관계 검정)과 변형된 시계열분석 방법을 이용하여 미얀마, 인도, 태국, 베트남 등 주요 쌀 수출국의 가격 변동에 대한 분석을 시도하였다. 분석에 활용된 자료는 2013 년 1 월부터 2017 년 12 월까지의 고품질 쌀 (5 % 쌀)과 저품질 쌀 (25 % 쌀)의 월별 국내 가격과 수출 가격이 사용되었다. 첫번째 단계로 Augmented Dickey Fuller (ADF) 테스트가 실행되었고 다음으로 Johansen 공적분 검정을 실시하여 국가 간의 연계성을 추정했다. 공적분 검정의 결과를 토대로 각 국가 간의 Granger 인과 관계를 분석 하였다. 그러나 본 연구는 국제 쌀 시장 (미얀마, 인도, 태국, 베트남)간에 장기간 가격 관계가 없다는 것을 발견했다. 또한 국제 쌀 시장 간의 밀접한 인과 관계를 발견하지 못했다. 따라서 전형적인 시계열 분석은 국제 쌀 시장 간의 밀접한 관계를 설명하지 못하는 것으로 해석하였다. 따라서 국가별 쌀 시장의 구조변화에 따른 국제 쌀 시장의 영향은 태국 고품질 쌀 가격을 제외한 모든 시장에서 일시적인 변화 만이 발견되었다. 본 논문의 추정결과를 바탕으로 국제 쌀 가격은 개별 국가의 국내 시장 변동성에 영향을 받는 것으로 판단하였다. 특히 미얀마 내수 시장과 수출가격이 다른 3 개의 주요 쌀 시장과의 장기적인 인과 관계를 발견하지 못하였기 때문에 미얀마의 쌀 시장은 다른 국가와는 다른 시스템을 보유하고 있는 것으로 판단된다. 그럼에도 불구하고 최신 정보 및 정확한 가격 데이터와 관련된 시장 정보는 본질적으로 정책결정에 필요하며, 국내 및 수출 시장뿐만 아니라 국제 시장과의 가격에 대한 연관성을 분석하기 위해 향후 연구에서는 고려될 필요가 있다. 마지막으로, 본 연구에서는 향후 쌀 가격의 변동성을 분석하기 위해 가용할 수 있는 자료를 토대로 고품질 쌀 가격을 예측하였다. Prices are the main incentives for the agricultural production and the signal for decision making process of all participants involved in the marketing process. The price of agricultural commodities performs the sign for production, welfare of consumers and determining the level of a nation’s export earnings. The study of price behavior is, therefore, critical to understand the market performance. This study was carried out to examine the rice price behavior across major rice exporting countries such as Myanmar, India, Thailand and Vietnam using a typical time-series process (stationary test, cointegration test, Granger Causality test), and the advanced time-series technique to detect and identify structural breaks and to produce forecasts. Monthly domestic and export rice price data in the case of high quality (5% broken rice) and low quality (25% broken rice) was used by covering from the period of January 2013 to December 2017. Firstly, stationary test for each price series was conducted by using Augmented Dickey Fuller(ADF) test. Then, Johansen cointegration test was performed to estimate the long run relationship between international rice markets. After that, the causal relationship between the price series was examined using Granger Causality test. However, in this study the existence of cointegration was not found among Myanmar domestic and export rice price in one side and between Myanmar and three major rice markets in another side, implying no long-run relationship between the markets. So, it can be concluded that Myanmar price and the international rice price are not related with each other. Also, we could not find any Granger causal relationship between Myanmar and international rice markets in high quality rice market. Thus, Myanmar and the international rice prices were independent and autonomous. Such results also reflected the unique rice market system in Myanmar. Finally, we detected and identified structural breaks to find volatility in individual international rice markets but we only found temporary changes for all markets except Thailand export high quality rice price. Based on the results we carefully conclude that rice price might follow an individual domestic market situation. Finally, we forecasted the future price for high quality rice and the result was reported within the range of highest and lowest for the future years.

Influence of Physical Environment, Composition of Vascular Plants and Habitat Structure on the Distribution of Mammal Species in Htamanthi Wildlife Sanctuary, Myanmar

Myo Min Latt 忠南大學校 大學院 2024 국내박사

In our planet, the plants are producers and the animals are consumers in food web. The environmental factors including natural and anthropogenic factors shapes the pattern of occurrence and diversity of both plant and animal. Therefore, the environmental factors and composition of different plant species surve as essential ecological niches for the distribution and diversity of animal species. Based on such scenario, this study was carried out in Htamanthi Wildlife Sanctuary (HWS), important corridor between Indo-Burma and Eastern Himalayan hotspot, located in the nother-western part of Myanmar. This study aimed to investigate the ecological relationship between physical environmental factors, composition of vascular plants and the distribution of large mammals. Vegetation survey was carried out by distributing 66 sample plots (sized 25×25m) with equal distance (5.7 km × 5.7 km) throughout the HWS. Regarding factors of the vascular plant, the name and diameter at breast height were recorded during vegetation survey, then habit, life form and types of forage were classified and hence the species richness and common diversity indices were calculated. The physical and chemical properties of soil were measured and tested. The elevation and slope were also measured as the topographic factors. The two types of raw datasets of mammal surveyed by CTS derived from four different projects and by RPS were applied. The mammals had been classified by the experts and trophic levels, body sizes, activeness and conservation status were categorized. Kruskal–Wallis test with Dunn–Bonferroni post hoc, redundancy analysis (RDA) ordination, negative binomial generalized lineral models (NBGLM), multivariate binomial logistic regression generalized linear models (MBLRGLM) and sigle species occupancy model (SSOM) with single season were applied. In this study, 371 vascular plant species belonging to 215 genera and 80 families were recorded. Five distinct forest communities were classified and their variation significantly associated with the soil Ca, Mg, clay proportion, soil hardness, and elevation. Sand, silt, stone and Ca are major factors affecting the variation of species richness, slope, stone, sand, silt, K, Mg explain the change of forest density and Hd, pH, stone, Total N are significant explaining the distribution of tree basal area. For diversity of mammal species, 65 species of mammals belonging to 43 genera and 16 families were recorded. This study suggested that combination of number of survey methods influenced on the detectable number of mammal species and multi-survey methods is better than only one survey method to monitor and detect the richness of mammal species and their distribution. The number of detectable events of large mammal species positively associated with number of camera trap locations, camera trap nights (CTNs), distance and duration of patrol and accessibility for RPS of the area. Among natural physical environmental features, the distance from main stream was the major important feature at community level but elevation, distance from old timber road and from villages were common for majority of mammal species at species level explaining the distribution of mammal species under different trophic level and body sized communities. Across the factors of ecological niche of floral forage plants composition, foliar, resin tree and subterranean forage plants were common factors at community levels but composition of foliar forage plant and resin tree were important positive factors at species level. Regarding the factors of forest structures, composition of shrub, bamboo (herb), seedling, sapling and canopy coverage were common important factors for distribution of mammals under different trophic communities but composition of bamboo (herb), seedling, sapling and trees were common factors for distribution of mammals under different body-sized communities. At species level, composition of tees, seedling, and bamboo were important positive factors for distribution of guar, leopard and tigers. In conclusion, identification of these community types and understanding the diversity levels and major factors influencing the community structure may play a key role in the planning, prioritization, and implementation of species and community conservation strategies amid the unpredictable impacts of climate change on forest community dynamics. Implementing wildlife survey by multi-methods is better approach in monitoring and assessment of wildlife in biodiversity conservation because different methods have different advantages. Main streams are main physical environmental factor to observe higher probability of occurrence of mammals in HWS. Gold mining may become a factor interrupting the distribution mammals because gold mines are located along the streams. Higher distribution of tiger and leopard in the area getting closer to villages should be concerned for the potential conflict with local people and their cattle. Dipterocarp spp. trees are very important proxy in predation which is very important biological interaction between preys and predictors in food web ecosystem. Composition of vascular plant species in terms of habit, life form and types of forage shapes the distribution of individual mammal species as well as play as important medium for prey and predators. These significant findings may support to understand ecological nich relationship, predation and interaction and ecological specialization. The relationship between animals and ecological niche of forage plant composition implicate in and provide the invaluable information for study of nutritional ecology and energy flow through food source availability.

Strategic policy decisions for the development of the port logistics sectors in Myanmar

Myo Nyein Aye 한국해양대학교 대학원 2013 국내박사

CNG/Diesel Dual Fuel Engine Combustion and Emissions Characteristics : An Experimental Analysis According to Common-rail Pressure and Fuel Mixing Ratio

MYO THANT ZIN 부경대학교 대학원 2020 국내석사

디젤 엔진은 일반적으로 발전기 및 이동식 차량 등 수많은 산업 분야에서 사용되고 있다. 그러나 디젤 엔진은 질소 산화물(NOx), 입자상 물질(PM), 이산화탄소(CO2) 등과 같은 환경오염 가스가 발생한다는 문제점이 있다. 이렇게 발생한 대기 오염 물질은 최근 환경 보호 위원회 등에 의해 엄격하게 규제가 되고 있다. 또한 한정된 화석연료 자원에 대하여 인류의 과도한 사용이 화석연료 자원 고갈을 가속화하고 있다. 따라서 디젤 엔진의 대체 연료 사용 및 개발이 연구자들 및 분야의 종사자들에게 있어 주요한 사항이다. 이러한 분야의 연구 및 발전을 위해, 본 논문은 주로 천연가스가 있는 이중 연료 디젤 엔진의 연소 특성에 대한 커먼레일 압력과 연료 혼합 비율의 변화에 대한 실험을 다룬다. 이 연구를 위해 커먼레일 4 행정 디젤 엔진을 이중 연료 모드로 운전하였으며, 디젤 연료 사용 시 엔진 특성과 비교하였다. 엔진 테스트는 적절한 분사 타이밍으로 11.7kW 및 23kW의 서로 다른 엔진 부하에서 3개의 커먼레일 압력(50MPa, 60MPa, 70MPa)에서 이루어졌으며, 천연가스 비율은 20%에서 90%까지를 사용하였다. 실험 결과는 열 방출 속도와 실린더 압력이 천연가스 비율의 증가와 비례 관계를 보여주었다. 또한, 모든 시험 조건에서 디젤 연료와 비교할 때 이중 연료 모드에서의 NOx와 CO 발생률이 감소하였다. 그러나 이중 연료 모드 결과로만 미루어볼 때, 커먼레일 압력이 증가함에 따라 CO는 감소하고 NOx는 증가했다. 요약하자면, 이중 연료 모드는 디젤 모드에 비해 NOx와 CO2 가스 배출이 적지만 HC와 CO의 발생률이 더 높게 나타났다. 또한, 커먼레일 압력이 증가하게 되면 HC와 CO 배출량은 감소하였다. 그러나 NOx와 CO2 배출량은 약간 증가하였다.