• Journal of KIISE
• /
• v.42 no.6
• /
• pp.748-754
• /
• 2015

This paper proposes a sequential state estimation model consisting of continuous and discrete variables, as a way of generalizing all discrete-state factorial HMM, and gives a design of gait motion model based on the idea. The discrete state variable implements a Markov chain that models the gait dynamics, and for each state of the Markov chain, we created a Gaussian process over the space of the continuous variable. The Markov chain controls the switching among Gaussian processes, each of which models the rotation or various views of a gait state. Then a particle filter-based algorithm is presented to give an approximate filtering solution. Given an input vector sequence presented over time, this finds a trajectory that follows a Gaussian process and occasionally switches to another dynamically. Experimental results show that the proposed model can provide a very intuitive interpretation of video-based gait into a sequence of poses and a sequence of posture states.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.7 no.6
• /
• pp.1309-1314
• /
• 2012

The SAMED filter is introduced as a wide class of multi-stage filters which encompass linear FIR and nonlinear order statistic filters. The output of SAMED filter is linear combination of sub-median outputs. In this paper, optimal SAMED filter is designed for images corrupted by various noise, and performance is analogized. The experimental result shows that the efficiency of each order of SAMED filters is depends on type of noise. It is shown that low order filters are effective in Gaussian environments but high order filters are effective in impulsive case. This result may be used to follow-up research on successive SAMED filters.

• Journal of Korea Game Society
• /
• v.9 no.5
• /
• pp.137-144
• /
• 2009

This paper predicts 6 facial regions that may have sharply different scattering properties, rendering the face more realistically based on their diffusion profiles. The scattering properties are acquired in the form of high dynamic range by photographing the pattern formed around an unit ray incident on facial skin. The acquired data are fitted to a 'linear combination of Gaussian functions', which well approximates the original diffusion profile of skin and has good characteristics as the filter. During the process, to prevent its solutions from converging into local minima, we take advantage of the genetic algorithm to set up the initial value. Each Gaussian term is applied to the irradiance map as a filter, expressing subsurface scattering effect. In this paper, to efficiently handle the maximum 12 Gaussian filterings, we make use of the parallel capacity of CUDA.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.19 no.5
• /
• pp.418-428
• /
• 2007

In the calculation of wave propagation by Boussinesq equation model, it is very common to experience numerical noises generated from nonlinear interaction and breaking wave occurrence, and the numerical solution is rapidly diverged unless the noises are properly controlled. A comparative study was here undertaken for the characteristics of three different lowpass filters (FFT filter, Gaussian filter and Shapiro filter) which are all designed to be applied to the interim results of numerical calculation. The numerical results obtained with application of respective filter techniques were compared with the results of an existing hydraulic experiment for the aspects of noise suppression, conservation of main signal and altering time. The results show that the Shapiro filter can be best applied with optimal choices of its element number, pass number and filtering tune interval. The combination of the number of filter element off, pass number of 50 or less, and application interval of 100 to 200 time steps generally showed good performance in both accuracy and efficiency of the numerical calculation.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.11 no.3
• /
• pp.471-478
• /
• 2007

The SA fillers encompass a large class of filters based on order statistics as veil as linear FIR filters. Using SA later structure, it is possible to design linear and non-linear filters under a unified framework. In this paper SA filters are applied to an image smoothing problem for mixed noise. Original image is contaminated by Gaussian and impulsive noise. Optimal SA filters are designed and applied to contaminated image. The experimental result shows that SA filters outperform linear FIR and ordering-based nonlinear filters.

• Proceedings of the Korea Institute of Convergence Signal Processing
• /
• 2000.12a
• /
• pp.133-136
• /
• 2000

에너지 차분 흥부 X선 화상으로부터 폐종류 음영을 검출하기 위한 필터를 예측해서 전문의의 진단보조 혹은 총합자동진단시스템의 구성 요소로서 필터가 발휘한 역할을 고려한 후, 구체적인 성능 평가 방법을 제안한다. 필터의 성능을 평가할 때 문턱값을 변화시킨 경우, 못보고 빠트린 비율과 잘못 본 비율의 변화를 ROC 곡선으로 나타내어 이것에 기초한 판단을 행하는 접근방법이 있다. 본 연구에서는 이와같은 평가에 대한 구체적인 예를 통하여 문제점을 명확히 하여 그 평가기준을 마련한다. 성능 평가를 행하기 위해 이용된 에너지 차분화상으로서, 임상 데이터 14개의 증예를 이용한다. 즉, 총 종류수 32개 한 개의 증예에 있어서 종류의 개수는 1~8개로 평균 2개, 1mm 당 샘플링 간격 5 픽셀, 비트 분해능 10 bit의 저압화상으로, 그 크기는 1760$\times$1760이다. 계산기 하드웨어의 제약으로부터 원화상을 가우스 함수로 평활화 해서 1/8로 축소한 화상으로 실험을 행한다. 상기 평가 절차에 따라, 이전에 개발된 다중해상도 ∇$^2$G 필터의 성능을 평가하고, 단일해상도 ∇$^2$G 필터와의 비교를 통해 그 성능이 우수함을 확인한다. 본 평가 방법은 화상진단 지원용 필터의 평가에 대해서 적용될 수 있을 것이다.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.1
• /
• pp.29-36
• /
• 2004

Generally, wavelet coefficients can be classified into two categories: large coefficients with much signal information and small coefficients with little signal component. This statistical characteristic of wavelet coefficient is approximated to Gaussian mixture model and efficiently applied to noise reduction. In this paper, we propose an image denoising method using mixture modeling of wavelet coefficients. Binary mask value is generated by proper threshold which classifies wavelet coefficients into two categories. Information of binary mask value is used to remove image noise. We also develope an enhancement method of mask value using morphological filter, and apply it to image denoising for improvement of the proposed method. Simulation results shows the proposed method have better PSNRs than those of the state of art denoising methods.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2017.10a
• /
• pp.473-476
• /
• 2017

As the society has developed into a digital information age society, a lot of electronic communication equipments are popularized. However, there are various causes of noise during signal transmission between communication devices. The noise generated in the communication system is a white noise that is distributed evenly in all frequency bands. This white noise causes system errors and lowers reliability. Therefore, in this paper, the existing Gaussian filter, Median filter, Alpha trimmed mean filter, and min/max filter for removing white noise are described and the characteristics and performance of each filter are compared with each other.

• Transactions of the Korean Society of Mechanical Engineers A
• /
• v.35 no.3
• /
• pp.289-297
• /
• 2011

The explicit finite element method is an essential tool for solving large problems with severe nonlinear characteristics, but its results can be difficult to interpret. In particular, it can be impossible to evaluate its acceleration responses because of severe discontinuity, extreme noise or aliasing. We suggest a new post-processing method for transient responses and their response spectra. We propose smoothing methods using a Gaussian kernel without in depth knowledge of the complex frequency characteristics; such methods are successfully used in the filtering of digital signals. This smoothing can be done by measuring the velocity results and monitoring the response spectra. Gaussian kernel smoothing gives a better smoothness and representation of the peak values than other approaches do. The floor response spectra can be derived using smoothed accelerations for the design.

• Journal of KIISE:Information Networking
• /
• v.29 no.5
• /
• pp.553-561
• /
• 2002

We present a novel service discipline, called linear service discipline, to serve multiple QoS queues sharing a resource and analyze its properties. The linear server makes the output traffic and the queueing dynamics of individual queues as a linear function of its input traffic. In particular, if input traffic is Gaussian, the distributions of queue length and output traffic are also Gaussian with their mean and variance being a function of input mean and input power spectrum (equivalently, autocorrelation function of input). Important QoS measures including buffer overflow probability and queueing delay distribution are also expressed as a function of input mean and input power spectrum. This study explores a new direction for network-wide traffic management based on linear system theories by letting us view the queueing process at each node as a linear filter.