• Journal of Welding and Joining
• /
• 제25권3호
• /
• pp.85-91
• /
• 2007

The conventional study on the laser surface transformation hardening has been carried out with a beam of the specified shape and uniform power-intensity distribution in order to ensure the uniformity of the hardening depth. Two types of beams - the circular gaussian beam and rectangular beam of the uniform power-intensity distribution were used in this study. we were supposed to optimize the process parameters and to compare the hardening results with two optics respectively. As a result, the hardness distribution of the hardened zone was similar in both cases and the hardened phase by the rectangular beam was denser than that by the circular gaussian beam.

• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• 제18권2호
• /
• pp.129-142
• /
• 2014

We propose a variational segmentation model based on statistical information of intensities in an image. The model consists of both a local region-based energy and a global region-based energy in order to handle misclassification which happens in a typical statistical variational model with an assumption that an image is a mixture of two Gaussian distributions. We find local ambiguous regions where misclassification might happen due to a small difference between two Gaussian distributions. Based on statistical information restricted to the local ambiguous regions, we design a local region-based energy in order to reduce the misclassification. We suggest an algorithm to avoid the difficulty of the Euler-Lagrange equations of the proposed variational model.

• 전자공학회논문지B
• /
• 제32B권1호
• /
• pp.127-135
• /
• 1995

The diffusion neural network forms a Gaussian distribution by transferring an excitation to the surround. A DOG(difference of two Gaussians) is obtained by the diffusion neural network. This type of the DOG, which can detect the intensity changes of an image, has the same shape as a LOG(Laplacian of a Gaussian:${\Delta}^2$G) and narrow band pass characteristics. In this paper we show that another type of the DOG which has a very narrow Gaussian for the excitatory and a very wide Gaussian for the inhibitory, can be formed by the diffusion process of this network, This type of the DOG has a wide band width in spatial frequency domain and can be used efficiently in detecting special type of edges.

• 한국정보통신학회논문지
• /
• 제18권3호
• /
• pp.735-741
• /
• 2014

에지는 영상에서 화소 간의 명암 차이가 큰 경우에 나타나며, 대상의 크기, 위치, 방향 등의 정보를 포함한다. 에지검출 방법에는 Sobel, Roberts, Laplacian, LoG(Laplacian of Gaussian) 등이 있으며, 이러한 방법들은 AWGN(additive white Gaussian noise)이 첨가된 영상에서 그 특성이 미흡하다. 따라서 이러한 특성을 개선하기 위하여 본 논문에서는 거리에 따른 가중치와 주변 화소의 평균에 의한 추정 마스크를 적용하는 알고리즘을 제안하였다. 그리고 제안한 방법의 성능을 확인하기 위하여 평가 척도는 처리 영상 및 PFOM(Pratt's figure of merit)을 사용하여 기존의 방법들과 비교하였다.

• 재활복지공학회논문지
• /
• 제5권1호
• /
• pp.95-101
• /
• 2011

본 논문에서는 의수환자의 일상생활을 고려한 1-자유도 동작을 손을 쥐고 폄으로 정의하고, 두 동작에 대한 근전도 패턴 분류를 위한 가우시안 혼합 모델 기반의 근전도 패턴 분류 알고리즘을 제안한다. 근전도 패턴 분류 알고리즘의 핵심이 되는 근전도 신호의 특징점 추출을 위하여 근전 신호의 진폭 특성을 고려하는 절대차분평균치(DAMV)와 평균절대값(MAV)을 사용한다. 또한 동작에 대한 근전 신호의 진폭 특성을 보다 명확히 구분하기 위하여 D_DAMV와 D_MAV를 제안한다. 본 논문에서는 4명의 성인남성을 대상으로 실험을 실시하였고, 두 동작에 대한 근전도 패턴의 정확한 분류 여부를 확인하였다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제17권8호
• /
• pp.1976-1995
• /
• 2023

The anchor-free object detector CenterNet regards the object as a center point and predicts it based on the Gaussian circle region. For each object's center point, CenterNet directly regresses the width and height of the objects and finally gets the boundary range of the objects. However, the critical range of the object's center point can not be accurately limited by using the Gaussian circle region to constrain the prediction region, resulting in many low-quality centers' predicted values. In addition, because of the large difference between the width and height of different objects, directly regressing the width and height will make the model difficult to converge and lose the intrinsic relationship between them, thereby reducing the stability and consistency of accuracy. For these problems, we proposed a center point prediction method based on the Gaussian elliptic region and a size component regression method based on the small solution space. First, we constructed a Gaussian ellipse region that can accurately predict the object's center point. Second, we recode the width and height of the objects, which significantly reduces the regression solution space and improves the convergence speed of the model. Finally, we jointly decode the predicted components, enhancing the internal relationship between the size components and improving the accuracy consistency. Experiments show that when using CenterNet as the improved baseline and Hourglass-104 as the backbone, on the MS COCO dataset, our improved model achieved 44.7%, which is 2.6% higher than the baseline.

• ETRI Journal
• /
• 제43권5호
• /
• pp.869-880
• /
• 2021

Herein, we estimate the direction of arrival (DOA) of non-Gaussian signals for nested arrays (NAs) by implementing the fourth-order difference co-array (FODC) and successive methods. In particular, considering the property of the fourth-order cumulant (FOC), we first construct the FODC of the NA, which can obtain O(N⁴) virtual elements using N physical sensors, whereas conventional FOC methods can only obtain O(N²) virtual elements. In addition, the closed-form expression of FODC is presented to verify the enhanced degrees of freedom (DOFs). Subsequently, we exploit the vectorized FOC (VFOC) matrix to match the FODC of the NA. Notably, the VFOC matrix is a single snapshot vector, and the initial DOA estimates can be obtained via the discrete Fourier transform method under the underdetermined correlation matrix condition, which utilizes the complete DOFs of the FODC. Finally, fine estimates are obtained through the spatial smoothing-Capon method with partial spectrum searching. Numerical simulation verifies the effectiveness and superiority of the proposed method.

• 스마트미디어저널
• /
• 제4권3호
• /
• pp.50-55
• /
• 2015

본 논문에서는 야간 영상 감시(night-time video surveillance)에 특화된 GMM(Gausssian mixture model)기반의 배경 모델링(background modeling)을 이용한 배경 차분(background subtraction)방법을 제안한다. 야간 영상에서는 낮 영상에 비해 배경과 객체의 구분이 뚜렷하지 않아 매우 흡사한 픽셀 값들을 이용하여 배경을 분리해야 한다. 이러한 문제점을 해결하기 위해 전처리 단계에서 조정된 범위의 히스토그램 스트레칭을 이용하여 입력 픽셀 값을 배경 모델링에 이로운 픽셀 값으로 변경해준다. 조정된 픽셀 값을 이용하여 가장 이상적인 배경을 찾기 위해 픽셀 단위로 GMM기반의 배경 모델링 방법을 적용한다. GMM을 기반으로 한 배경모델링 방법에서는 새로운 픽셀 값이 입력되었을 때 어떤 가우시안에도 속하지 않는다면 가장 낮은 가중치를 가진 가우시안 분포를 제거함으로써 이전의 축적된 배경의 정보를 무시하는 결과를 낳게 된다. 따라서 본 논문에서는 낮은 가중치의 가우시안을 제거하는 대신 기존 가우시안의 평균과 입력된 픽셀 값의 차를 이용하여 새로운 평균에 적용함으로써 기존의 쌓여진 정보를 고려한다. 실험결과 제안된 배경 모델링 방법이 기존 방법의 이점을 유지하면서 야간 영상 감지에 특화된 배경 차분 결과를 보였다.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2002년도 하계종합학술대회 논문집(4)
• /
• pp.219-222
• /
• 2002

In this paper, we propose a new blind image watermarking method in wavelet domain. It is necessary to find out watermark insertion location in blind watermark. We use horizontal and vertical difference of LL components to select watermark insertion location, because increment or decrement of successive components is rarely changed in LL band. A pseudo-random sequence is used as a watermark. Experimental results show that the proposed method is robust to various kinds of attacks such as JPEG lossy compression, averaging, median filtering, resizing, histogram equalization, and additive Gaussian noise.

• 한국통신학회:학술대회논문집
• /
• 한국통신학회 1986년도 추계학술발표회 논문집
• /
• pp.182-185
• /
• 1986

In this paper, the four algorithm which obtain 2D displacement vector are proposed. In corwocutive difference pictures, the characteristics of up DP boundary and region are discussed and we estimate displacement vector using the DP boundary and region, Finally, the performance of proposed algorithm for gaussian noisy image which generated by computer are discussed.