• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2004.10a
• /
• pp.543-546
• /
• 2004

Bishop과 Nabney에 의해 소개된 기존의 혼합 밀도 네트워크(Mixture Density Network)에서는 조건부 확률밀도 함수의 매개변수들(parameters)이 하나의 MLP(multi-layer perceptron)의 출력 벡터로 주어진다. 최근에는 변형된 혼합 밀도 네트워크(Modified Mixture Density Network)라고 하는 이름으로 조건부 확률밀도 함수의 선분포(priors), 조건부 평균(conditional means), 그리고 공분산(covariances) 등이 각각 독립적인 MLP의 출력벡터로 주어지는 경우를 다룬 연구가 보고된 바 있다. 본 논문에서는 조건부 평균이 입력에 관해 선형인 경우를 위한 버전에 대한 이론과 매트랩 프로그램 개발 및 적용을 다룬다. 본 논문에서는 우선 일반적인 혼합 밀도 네트워크에 대해 간단히 설명하고, 혼합 밀도 네트워크의 출력인 다층 퍼셉트론의 매개변수를 각각 다른 다층 퍼셉트론에서 학습시키는 변형된 혼합 밀도 네트워크를 설명한 후, 각각 다른 다층 퍼셉트론을 통해 매개변수를 얻는 것은 동일하나 평균값은 선형함수를 통해 얻는 혼합 밀도 네트워크 버전을 소개한다. 그리고, 모의실험을 통하여 이러한 혼합 밀도 네트워크를의 적용가능성에 대해 알아본다.

• Annual Conference of KIPS
• /
• 2009.04a
• /
• pp.94-97
• /
• 2009

본 논문에서는 실제 도로 환경에서의 실시간 차선 인식 방법을 제안한다. 전방주시카메라를 활용하여 촬영한 입력영상으로부터 도로영역에 해당하는 관심영역을 추출하고 반복적인 평균 명도를 측정하여 이진화함으로써 차선 특징을 검출하고 YCbCr 변환한 영상에 대한 실험 임계값을 적용하여 중앙선의 특징을 검출하였다. 이에 Canny 알고리즘을 이용한 에지 추출로 허프 변환시의 작업량을 최소화하였으며 허프 변환하여 얻은 차선 후보군으로부터 각도를 기반으로 필터링하여 통계적으로 우선순위가 높은 선분을 차선으로 인식하였다. 또한 실제 도로 환경에서 수집한 동영상으로 실험한 결과 강건한 차선 인식률을 보였다.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.10B
• /
• pp.994-1003
• /
• 2009

For Automatic Meter Reading System, good topology of check machines, concentrators, and repeaters in client field is important. Steiner Minimum Tree is a minimum cost tree connecting all given nodes with introducing Steiner points. In this paper, an efficient mechanism allocating and connecting check machines, concentrators and repeaters which are essential elements in automatic meter reading system is proposed, which conducts repeated applications of building approximate Minimum Steiner Trees. In the mechanism, input nodes and Steiner points might correspond to check machine, concentrators or repeaters and edges might do to the connections between them. Therefore, through suitable conversions and processes of them, an efficient network for automatic meter reading system with both wired and wireless communication techniques could be constructed. In our experiment, for 1000 input nodes and 200 max connections per node, the proposed mechanism shortened the length of produced network by 19.1% comparing with the length of Minimum Spanning Tree built by Prim's algorithm.

• Journal of KIISE:Software and Applications
• /
• v.27 no.6
• /
• pp.583-594
• /
• 2000

The thinning process which is commonly used in extracting skeletons from handwritten Hangul characters has a problem of distorting the original pattern shapes. This paper proposes a method of skeleton extraction using a shape decomposition algorithm. We decompose the character pattern into a set of near convex parts using a shape decomposition algorithm. From the shape-decomposed pattern, we detect the joint parts and extract the skeletons from the parts incident to the joint parts. Then the skeletons not incident to the joint parts are extracted. Finally, the process of skeleton extension is performed to ensure the connectivity. We setup five criteria for the comparison of quality of skeletons extracted by our method and the thinning based method. The comparison shows the superiority of our method in terms of several criteria.

• Journal of Advanced Navigation Technology
• /
• v.11 no.3
• /
• pp.343-349
• /
• 2007

This paper addressed a recognition method of land registration map based on fuzzy inference scheme, which is able to solve the time complexity problem of typical method [Fig. 2]. Not only line color, thickness but also number, character are used as a fuzzy input parameter. It concentrated on generation of fuzzy association map, and useful informations are extracted result from fuzzy inference. These results are precedent process for estimating the construction space and restoring 3D automatic modeling. It can also utilize to the internet service acceleration propulsion business such as u-Gov based land registration service.

• Proceedings of the Korean Information Science Society Conference
• /
• 2002.10d
• /
• pp.148-150
• /
• 2002

본 논문은 모바일 핸드셋이라는 제한적인 디스플레이 환경과 입력환경에서 필요한 라우트맵의 시각화에 따른 문제점과 그 해결방안을 제시한 논문이다. 자동차 네비게이션 시스템에서 널리 사용되는 경로표시 방법은 일반적으로 노트북이나, PDA, 전용 디스플레이 시스템등에서 구현되어 있으나, 모바일 핸드셋과 같이 제한된 디스플레이 환경에서의 경로표시 방법은 현재까지 많은 연구가 이루어지지 않고 있다. 본 논문은 인간의 인식능력이 임의의 경로를 회전방향점(turning point)의 연속으로 인식한다는 점에 착안하여 원래의 경로가 가진 축적을 변형하여 선분과 회전방향점으로 경로를 표시하는 라인 드라이브 시스템을 모바일 환경의 경로표시 방법으로 사용하였다. 본 논문에서 구현한 시스템은 GIS의 벡터 정보를 가지고 있는 서버와 Brew 플랫폼을 지원하는 모바일 핸드셋으로 구성된다. GIS 서버는 프로토타입 정보를 추출하여 모바일 핸드셋으로 전송하고, 이러한 정보는 모바일 핸드셋의 Brew 플랫폼에서 간략화, 도식화 과정을 거쳐서 시각화된다. 본 논문은 제한된 디스플레이 환경의 경로 확대표시를 위하여 계층별 경로 시각화 방법을 사용하여 가독성을 높이고 사용상의 편리함을 추구한다.

• Journal of Korea Game Society
• /
• v.14 no.2
• /
• pp.55-66
• /
• 2014

Image warping is a technology to transform input images to be suitable for given conditions and has been recently utilized in changing face shape of characters in the field of movies or animation. Mesh warping which is one of warping methods that change shapes based on the features of face forms warping images by forming rectangular mesh groups around the eyes, nose, and mouth and matching them 1:1. This method has a problem in the resultant images are distorted in the segments of boundaries between meshes when there are errors in mesh control points or when meshes have been formed as many small area meshes. This study proposes a triangle based image interpolation technique to minimize the occurrence of errors in the process of forming natural warping images of face and process accurate results with a small amount of arithmetic operation and a short time. First, feature points that represent the face are found and these points are connected to form basic triangle meshes. The fact that the proposed method can reduce errors occurring in the process of warping while reducing the amount of arithmetic operation and time is shown through experiments.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.10 no.12
• /
• pp.2315-2320
• /
• 2006

A real-time frame-layer rate control algorithm with a token bucket traffic shaper is proposed for low bit rate video coding. The proposed rate control method uses a non-iterative optimization method for low computational complexity, and performs bit allocation at the frame level to minimize the average distortion over an entire sequence as well as variations in distortion between frames. In order to reduce the quality fluctuation, we use a sliding window scheme which does not require the pre-analysis process. Therefore, the proposed algorithm does not produce time delay from encoding, and is suitable for real-time low-complexity video encoder. Experimental results indicate that the proposed control method provides better visual and PSNR performances than the existing rate control method.

• Journal of the Korea Computer Graphics Society
• /
• v.23 no.3
• /
• pp.95-103
• /
• 2017

Horizon correction is a crucial stage for image composition enhancement. In this paper, we propose a deep learning based method for estimating the slanted angle of a photograph and correcting it. To estimate and correct the horizon direction, existing methods use hand-crafted low-level features such as lines, planes, and gradient distributions. However, these methods may not work well on the images that contain no lines or planes. To tackle this limitation and robustly estimate the slanted angle, we propose a convolutional neural network (CNN) based method to estimate the slanted angle by learning more generic features using a huge dataset. In addition, we utilize multiple adaptive spatial pooling layers to extract multi-scale image features for better performance. In the experimental results, we show our CNN-based approach robustly and accurately estimates the slanted angle of an image regardless of the image content, even if the image contains no lines or planes at all.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.14 no.7
• /
• pp.847-851
• /
• 2004

In the original mixture density network(MDN), which was introduced by Bishop and Nabney, the parameters of the conditional probability density function are represented by the output vector of a single multi-layer perceptron. Among the recent modification of the MDNs, there is the so-called modified mixture density network, in which each of the priors, conditional means, and covariances is represented via an independent multi-layer perceptron. In this paper, we consider a further simplification of the modified MDN, in which the conditional means are linear with respect to the input variable together with the development of the MATLAB program for the simplification. In this paper, we first briefly review the original mixture density network, then we also review the modified mixture density network in which independent multi-layer perceptrons play an important role in the learning for the parameters of the conditional probability, and finally present a further modification so that the conditional means are linear in the input. The applicability of the presented method is shown via an illustrative simulation example.