관리도기법을 이용한 인간 움직임의 응급상황 발생에 대한 탐지시스템 개발

조완현,손영숙,박순영,김희정 한국자료분석학회 2006 Journal of the Korean Data Analysis Society Vol.8 No.1

This paper proposes a discrimination system of human motions using acceleration sensors and CUSUM control chart. Our approach uses motion features that are represented as three kinds of waves in two dimensional spaces. To express exactly a pattern of each action, we drive the linear predictor coefficients from 3D sensor data and use these values as feature vectors. Next, to recognize each action, we use Bayes Classifier and Neural Network Classifier and we use CUSUM control chart technique to monitoring the human real-life. Experimental results show that our algorithm is very accurate and also is robust to the recognition of a change of each activity in performance of various physical styles of human actions. 본 논문에서는 여러 가지 인간의 움직임형태를 가속센서를 이용하여 측정하고, 베이스의 패턴인식 방법과 관리도 기법을 이용하여 인간의 일상 활동 중에서 갑작스런 응급상황이 일어났는지를 탐지할 수 있는 인식시스템을 제안한다. 본 연구에서는 인간의 동작들을 삼차원 가속센서로부터 측정되는 파동형태의 특징벡터로 표현하고, 이들 측정값으로부터 패턴인식에 적절한 선형예측계수와 짧은 시간 에너지 값들을 추출하여 패턴특징벡터로 사용하였다. 또한 패턴을 인식하는 분류기로는 베이스 분류기를 이용하여 모든 동작들이 정상상태의 움직임인지 또는 비 정상상태의 움직임인지를 판별하였다. 다음으로 누적합 관리도 방법을 사용하여 실시간으로 인간의 동작상태를 감시하여 이상상태가 발생하였을 때 이것을 신속하게 탐색할 수 있는 탐색시스템을 제안한다. 실험 결과를 통하여 제안된 알고리즘이 여러 사람들의 다양한 움직임의 형태를 비교적 정확하게 잘 인식할 수 있음을 보였고, 또한 베르누이 누적합 관리도 방법이 인간행동방식에 대한 이상적인 변화의 발생을 정확하게 탐지하고 있음을 알 수 있었다주요용어 : 인간행동유형, 가속센서, 선형예측계수, 짧은시간 에너지계수, 베이스 분류기, 오분류행렬, 베르누이 누적합 관리도.

Biplot 이론과 타이어 제조공학에의 응용

조완현 한국통계학회 1996 응용통계연구 Vol.9 No.2

일반적으로 타이어 제조공학에 있어서 측정되는 자료는 여러개의 설계인자와 성능 특성치가 사용된 다변량 자료행렬로 주어지는데, 이러한 자료행렬의 중요한 특성중의 하나는 각 반응값들이 서로 다른 것들과 높은 상관관곌르 가질 수 있다는 것이다. 따라서 본 연구는 이러한 복잡한 성격을 갖는 자료행렬의 분석에 적합한 biplot 작성의 수학적 이론을 알아보고, 또한 각 변수들의 구조적 특성 및 내재한 상호 관련성을 다변량 분산분석 biplot과 다반응치 희귀모형 biplot을 이용하여 포괄적으로 고찰하였다. It is essential in modern industry that quality and procuctivity are improved continuously. To accomplish this purpose, quality control must be maintained in all parts of a company. Recently, some tire manufacture companies are beginning to show interest in quality control. They have tried to achive some results through the statistical analysis for the experimental data which has accumulated up to now and then they strive to determine the structural relationship between the design factors in tire construction and tire performance characteristics. The measurement data obtained from the construction engineering is given in multivariate form owing to the various properties found in tire design components as wll as in performance. Also it may be existed the relationship among the multimple response variables. Thus we proposes the use of the biplot graphical display as an analytic tool of data matrices with complex respects. The proposed biplots are also availalbe to understand both the underlying structure of the data and the roles played by the different components. In particular, we consider the matter of how best to use the biplots in the maltivariate analysis of variance and multiple response data. Finally we apply this method to analyze the actual data.

Exploring Environmental Factors Affecting Strawberry Yield Using Pattern Recognition Techniques

조완현,박유하,나명환,최돈우 한국인터넷정보학회 2019 인터넷정보학회논문지 Vol.20 No.1

This paper investigates the importance of various environmental factors that have a strong influence on strawberry yields grown in greenhouse using the pattern recognition methods. The environmental factors influencing the production of strawberries were six factors such as average inside temperature, average inside humidity, average CO₂ level, average soil temperature, cumulative solar radiation, and average illumination. The results of analyzing the observed data using Dynamic Time Warping (DTW) showed that the most significant factor influencing the strawberry production was average soil temperature, average inside humidity, and cumulative solar radiation. Second, the results of analyzing the observed data using Multidimensional Scaling (MDS) showed that the most influential factors on the strawberry yields, such as average CO2 level, average inside humidity, and average illumination were differently given for each farms. However, these results are based on the distance in 3D space and can be deduced from the fact that there is not a large difference between these distances. Therefore, in order to increase the harvest of strawberries cultivated in the farms, it is necessary to manage the environmental factors such as thoroughly controlling the humidity and maintaining the concentration of CO₂ constantly by ventilation of the greenhouse.

Various Color Models and Color Feature Indexes to Predict Fruit Ripeness Stages

조완현,나명환,김상균 한국자료분석학회 2018 Journal of the Korean Data Analysis Society Vol.20 No.5

In general, farmers who grow fruits such as strawberry or tomato in greenhouses are very eager to predict the optimal harvesting time for fruits in advance. Therefore, in order to solve these farmers' needs, this paper suggests the appropriate color indexes representing a ripeness of fruit images which are essential to develop a system that can automatically determine the degree of maturity of fruit. First, we consider the definitions and properties of three color models such as CIE XYZ, CIE LAB, and CIE LUV that are commonly used to represent fruit images. Second, we examine the color components and the extracting method of color feature vectors that is most suitable to accurately determine the ripening stage of the fruit. From the experimental results, the color components that can accurately distinguish the stages of ripening of strawberry were Y component in XYZ color model, A component in LAB color model, and V component in LUV color model. Also, it can be seen that the average, skewness and kurtosis of color values can be used as color feature vectors effective for determining the degree of ripening of strawberry. Therefore, the results of our study are expected to be useful to develop a system that can automatically determine the harvesting time of fruits.

Human Action Recognition Using a Hybrid Method of a Generative Model and a Discriminative Model

조완현 한국자료분석학회 2016 Journal of the Korean Data Analysis Society Vol.18 No.5

In this paper, we proposed a hybrid system of two powerful machine learning schemes, namely the generative model HMM (hidden Markov model) and the discriminative model SVM (support vector machines), which can concertedly enhance the accuracy of human action recognition. The basic idea of our hybrid method is that the temporal characteristics of the sequence data are modeled by HMM state transitions and both the SVM model and Platt’s method obtained through the preprocessing are used to compute the emission probabilities of observations given at each state in HMM. In order to evaluate the performance of the proposed model, we have conducted an experiment that is able to compare our hybrid model with other classical methods using a KTH dataset. From our experimental results, we note that our hybrid approach shows better or poor performance than existing methods in various fields of human action.

집락분석법에 있어서 비유사도와 계층적 응집법의 관계에 관한 연구

조완현 한국통계학회 1992 응용통계연구 Vol.5 No.2

본논문은 지금까지 집락분석방법에서 많이 사용되는 유사도 또는 비유사도의 정의 및 이들의 수학적 성질을 알아보고, 또한 이들에 의해서 생성되는 비유사도 행렬을 이용하여 계층적 집락분석을 실시하였다. 이 경우에 가정된 초기의 집락 구조를 정확하게 잘 재생시킬 수 있는 비유사도의 측정방법과 계층적 응집법의 상호 관계를 질적자료와 양적자료 각각에 대하여 고찰하고, 이들에 관련된 시뮬레이션 결과를 제시하였다. In this paper we consider the definition and mathematical properties of similarity or dissimilarity which have often used in clust analysis, and we apply a hierarchical agglomerative cluster algorithm to a dissimilarity metrx generated by these distance. Here we investigate the effect of relation between distance function and cluster algorithm on the retrieval ability of natural clusters. We present an empirical results for qualitative data as well as quantitative data.

평균이동절차와 결정적 어닐링 EM 알고리즘을이용한 자율적 칼라영상분할

조완현,박종현,이명은,박순영 한국자료분석학회 2005 Journal of the Korean Data Analysis Society Vol.7 No.5

We present an unsupervised segmentation algorithm combining the mean shift procedure and deterministic annealing Expectation Maximization (DAEM) called MS-DAEM algorithm. We use the mean shift procedure to determine the number of components in a mixture model and to detect their modes of each mixture component. Next, we have adopted the Gaussian mixture model (GMM) to represent the estimate the parameters of the GMM that represents the multi-colored objects statistically. The experimental results show that the mean shift part of the proposed MS-DAEM algorithm is efficient to determine the number of components and initial modes of each component in mixture models. And also it shows that the DAEM part provides a global optimal solution for the parameter estimation in a mixture model and the natural color images are segmented efficiently by using the GMM with components estimated by MS-DAEM algorithm. 본 연구에서 우리는 평균이동절차(Mean Shift Procedure)와 결정적 어닐링 EM(Deterministic Annealing EM) 알고리즘을 결합한 새로운 칼라영상분할 알고리즘을 제안한다. 먼저 칼라영상을 혼합모형으로 표현할 때 혼합모형의 수와 각 모형들의 최대값(Mode)의 위치를 결정하는데 평균이동절차를 사용한다. 그리고 혼합모형으로 다변량 가우시안 혼합모형(Gaussian Mixture Model)을 가정하고, 이들 가우시안 확률모형의 모수인 평균벡터와 공분산행렬을 추정하는데 결정적 어닐링 EM 알고리즘을 사용하였다. 실험 결과들은 제안된 알고리즘의 전반 부분인 평균이동절차가 혼합모형의 각 성분의 수와 최대값의 위치를 결정하는데 매우 효율적인 방법임을 보여주고 있으며 또한 알고리즘의 후반 부분인 결정적 어닐링 EM 알고리즘이 혼합모형의 모수들에 대한 전역적 최적 추정량을 제공할 수 있는 최적의 추정방법임을 보이고 있다. 끝으로 다양한 칼라영상분할 실험을 통하여 새롭게 제안된 알고리즘이 칼라영상분할을 자율적으로 수행할 수 있는 가장 효율적인 알고리즘임을 보였다.

분산 베이지안 추론방법과 가우시안 확률과정 분류모형을 이용한 다차원 자료의 분류

조완현 한국자료분석학회 2015 Journal of the Korean Data Analysis Society Vol.17 No.6

In this paper, we present the variational Bayesian Gaussian process classification model for the classification of multidimensional data. Here, we have considered an efficient computational method that can be used to obtain the approximate posterior distributions for latent variables and various parameters need to define the multi-classes Gaussian process classification model. We have first induced the approximate posterior distributions for parameters and latent function by using variational Bayesian inference method and important sampling technique, and next we derived a predictive distribution of latent function used at the classification of new sample. We have conducted experiments that can verify the performance of our model for two multidimensional dataset. Experimental results show the fact that our method may classify efficiently the multidimensional data. 최근에 새롭게 제안되고 있는 가우시안 확률과정 분류모형에 대한 통계적 추론절차로 분산 베이지안 추론방법을 고찰하였고, 이모형을 이용하여 실제 다차원 자료를 분류할 수 있는 활용방안을 제시하였다. 먼저 우리는 다차원 자료의 분류범주를 나타내는 잠재변량에 대한 사전분포로 가우시안 확률과정을 가정하였고, 관측자료에 우도함수로 로지스틱 함수를 고려하여 잠재변량에 대한 사후분포를 유도하는데 근사적 베이지안 추론방법을 고려하였다. 이때 분류모형을 정의하기 위해서 필요한 잠재변량과 다양한 모수들에 대한 사후분포를 유도하는데 분산 베이지안 추론방법과 중요샘플기법을 사용하였고, 또한 새로운 관측 자료를 분류하는데 사용될 수 있는 잠재변량에 대한 예측분포를 유도하였다. 제안된 모형의 성능을 평가하기 위해서 두 종류의 다변량 자료를 사용하여 실험을 실시하였다. 실험결과들은 제안된 방법이 이들 자료들을 효율적으로 잘 분류함을 보여주고 있다.

Edge-Preserving Denoising Method Using Variation Approach and Image Gradient Distribution

조완현,서성채,유진호 한국자료분석학회 2013 Journal of the Korean Data Analysis Society Vol.15 No.6

This paper proposed an image denoising technique that can enhance the quality of image by using a variational approach and image gradient distribution. First, in order to remove the noise, we consider the variational approach for the energy functional that satisfies an edge-preserving regularization property. Here, we propose a new variational functional that can be implemented by adding a new gradient distribution term in a given energy functional that locally controls the extent of denoising over image regions according to their gradient magnitudes. And by using the fundamental lemma for the calculus of variations, we derive the Euler-Lagrange equation for true image that can achieve the minimum of a devised functional. Next, we considered the procedure that this equation can be solved by using a gradient decent method, which is one of the dynamic approximation techniques. Through various experiments, we can demonstrate that the proposed method can preserve the edges while removing noise better than existing techniques.

Detection and Recognition of Moving Objects using the Moving Mask Method and the Hidden Markov Model

조완현,김선월,안국동 한국자료분석학회 2011 Journal of the Korean Data Analysis Society Vol.13 No.1

This paper proposes a new detection and recognition method for moving objects that uses the moving mask method and the hidden Markov model. First, we apply the concept of entropy to convert the pixel value in the image domain into the amount of energy change in the entropy domain. Second, we find the coarse region of moving objects by automatically generating the moving mask image from the inter-frame difference between two transformed entropy images. Third, we use the discrete wavelet transformation technique to extract proper feature vectors from the detected mask image. Fourth, we use the hidden Markov model to accurately recognize moving objects. The results indicate that our proposed method can effectively and accurately detect and recognize moving objects in image sequences.