• 한국멀티미디어학회논문지
• /
• 제4권6호
• /
• pp.518-523
• /
• 2001

미세한 차이를 고감도 식별하는 카오스 이론의 프랙탈 차원과 에농 시스템에서 발생하는 이상한 끌개(Strange Attractor)를 이용하여 문자 특징을 추출, 문자 인식에 적용하는 새로운 방법을 제안함으로써 일반문자 뿐만 아니라, 문자들의 유사성에 의해 오인식되는 혼동 문자를 프랙탈 차원 해석에 의해 해소하는 고정도 문자 인식 시스템을 구현한다. 먼저, 문자 영상으로부터 문자의 고유 성질을 나타내는 망 특징 및 투영 특징, 교차거리 특징 등을 1차 구한 후, 이들 특징을 시계열 데이터로 변환한 다음, 이를 본 논문에서 제안한 수정된 에농 시스템을 이용하여, KS C 5601 표준 한글 2,350자에 대 한 각각의 문자 어트랙터를 재구성한다. 다음 단계에서는 개별 문자 어트랙터의 혼돈도를 분석하기 위해 각각의 문자에 대하여, 프랙탈 차원을 나타내는 정보 차원값(Box-counting Dimension, Natural Measure, Information Bit, Information Dimension)을 계산하여 문자 영상의 최종 특징을 구한다. 실험결과 한글 2,350자에 대하여 99.49%은 분류율을 나타내어 제안된 방법의 유효성을 보였다.

• Current Optics and Photonics
• /
• 제2권1호
• /
• pp.69-78
• /
• 2018

This paper presents a precise edge detection algorithm for the critical dimension (CD) measurement of a Thin-Film Transistor Liquid-Crystal Display (TFT-LCD) pattern. The sigmoid surface function is proposed to model the blurred step edge. This model can simultaneously find the position and geometry of the edge precisely. The nonlinear least squares fitting method (Levenberg-Marquardt method) is used to model the image intensity distribution into the proposed sigmoid blurred edge model. The suggested algorithm is verified by comparing the CD measurement repeatability from high-magnified blurry and noisy TFT-LCD images with those from the previous Laplacian of Gaussian (LoG) based sub-pixel edge detection algorithm and error function fitting method. The proposed fitting-based edge detection algorithm produces more precise results than the previous method. The suggested algorithm can be applied to in-line precision CD measurement for high-resolution display devices.

• 한국전기전자재료학회:학술대회논문집
• /
• 한국전기전자재료학회 2008년도 하계학술대회 논문집 Vol.9
• /
• pp.74-75
• /
• 2008

This work has suggested corrugation beam as a new structure for mechanical resonators. Micro beam resonators based on 3C-SiC films which have two side corrugations along the length of beams were simulated by finite-element modeling and compared to a flat rectangular beam with the same dimension. With the dimension of $36\times12\times0.5{\mu}m^3$, the flat cantilever has resonant frequency of 746 kHz. Meanwhile, this frequency reaches 1.252 MHz with the corrugated cantilever which has the same dimension with flat type but corrugation width of $6{\mu}m$ and depth of $0.4{\mu}m$. It is expected that mechanical resonators with corrugations will be very helpful for the research of sensing devices with high-resolution, high-performance oscillators and filters in wireless communications as well as measurement in basic physics.

• Industrial Engineering and Management Systems
• /
• 제13권4호
• /
• pp.454-462
• /
• 2014

Retention of possible churning customer is one of the most important issues in customer relationship management, so companies try to predict churn customers using their large-scale high-dimensional data. This study focuses on dealing with large data sets by reducing the dimensionality. By using six different dimension reduction methods-Principal Component Analysis (PCA), factor analysis (FA), locally linear embedding (LLE), local tangent space alignment (LTSA), locally preserving projections (LPP), and deep auto-encoder-our experiments apply each dimension reduction method to the training data, build a classification model using the mapped data and then measure the performance using hit rate to compare the dimension reduction methods. In the result, PCA shows good performance despite its simplicity, and the deep auto-encoder gives the best overall performance. These results can be explained by the characteristics of the churn prediction data that is highly correlated and overlapped over the classes. We also proposed a simple out-of-sample extension method for the nonlinear dimension reduction methods, LLE and LTSA, utilizing the characteristic of the data.

• Structural Engineering and Mechanics
• /
• 제41권5호
• /
• pp.663-673
• /
• 2012

A number of mode shape-based structure damage identification methods have been verified by numerical simulations or experiments for on-line structure health monitoring (SHM). However, many of them need a baseline mode shape generated by the healthy structure serving as a reference to identify damages. Otherwise these methods can hardly perform well when multiple cracks conditions occur. So it is important to solve the problems above. By aid of the fractal dimension method (FD), Qiao and Wang proposed a generalized fractal dimension (GFD) to detect the delamination damage. As a modification of GFD, Qiao and Cao proposed the approximate waveform capacity dimension (AWCD) technique to simplify the calculation of fractal and overcome the false peak appearing in the high mode shapes. Based on their valued work, this paper combined and applied the AWCD method and curvature mode shape data to detect multiple damages in beam. In the end, the identification properties of the AWCD for multiple damages have been verified by groups of Monte Carlo simulations and experiments.

• Journal of Electrical Engineering and Technology
• /
• 제13권5호
• /
• pp.2099-2106
• /
• 2018

This study presents a novel approach of discriminative feature vectors based on manifold learning using nonlinear dimension reduction (DR) technique to improve loss function, and combine with the Adversarial examples to regularize the object function for image classification. The traditional convolutional neural networks (CNN) with many new regularization approach has been successfully used for image classification tasks, and it achieved good results, hence it costs a lot of Calculated spacing and timing. Significantly, distrinct from traditional CNN, we discriminate the feature vectors for objects without empirically-tuned parameter, these Discriminative features intend to remain the lower-dimensional relationship corresponding high-dimension manifold after projecting the image feature vectors from high-dimension to lower-dimension, and we optimize the constrains of the preserving local features based on manifold, which narrow the mapped feature information from the same class and push different class away. Using Adversarial examples, improved loss function with additional regularization term intends to boost the Robustness and generalization of neural network. experimental results indicate that the approach based on discriminative feature of manifold learning is not only valid, but also more efficient in image classification tasks. Furthermore, the proposed approach achieves competitive classification performances for three benchmark datasets : MNIST, CIFAR-10, SVHN.

• 한국초전도ㆍ저온공학회논문지
• /
• 제4권2호
• /
• pp.11-15
• /
• 2002

High speed probe for measurement of sin91e flux quantum circuits is comprised of coaxial cables and microstrip lines in order to carry high speed signals without loss. For the impedance matching between coaxial cable and microstrip line, we have determined the dimension of the microstrip line with 50${\Omega}$ impedance by simulation and then have investigated the effect of line width and cross-sectional shape of signal line, dielectric material, thickness of soldering lead at the coaxial-to-microstrip transition Point, and the an91c between dielectric material and end part of the signal line on the characteristics of signal transmission of the microstrip line. From the simulation, we have found that these all parameter's had influenced on the characteristic of signal transmission on the microstrip line and should be reflected in fabricating high speed probe, We have also determined the dimension of coplanar waveguide to fabricate testing sample for performance test of high speed probe.

• 대한가정학회지
• /
• 제28권3호
• /
• pp.13-24
• /
• 1990

The main purpose of this study are as follows ; 1) To examine closely the effect of personality dimension on brand image. 2) To investigate the difference of variables about brand image according to the characteristics of the population statistics and draw the strategies of marketing for our wear enterprises. A 300 Samples were selected from female in Seoul and the investigation was conducted during 21 days, from 1998. 9. 21 to 1988. 10. 11. As for survey methozs, the personality dimension test developed by Eysenk was adopted. To measure the brand image, the adjectives of the semantic differentia scale developed by Malhotra and adjective that has been used in various were image analysis were adopted. The data were analysed using the statistical technic of Correlation Coefficient, F-test, and X2 test. The Results obtained from this study were as follows. 1. There were partially significant relationships between adult female's four subordinate variables of the personality dimension and preferred brand image on Women's ready-made wear. 1) The people having a high Psychoticism tendency preferred "individual" image and less preferred "practical" image than the people of low Psychoticism. 2) The people having a high extraversion tendency preferred "bold", "aged" image and less preferred "feminine", "practical" image. 3) The unstable female having a high neuroticism tendency preferred "abscure" image and less preferred "Practical" "gaudy", "Open hearted" image. 4) The people having a high lie tendency perferred "intricate", "classical" image and less preferred "bold", "citified", "incongruous" image. 2. There were partially significant differences in adult female's preferred brand image on women's ready made wear according to the characteristics of the population statistics.

• 자연과학논문집
• /
• 제4권
• /
• pp.191-220
• /
• 1991

Human Visual System(HVS)의 특성과 image의 textural regions의 roughness을 이용하여 image segmentation을 행하여 high compression에서도 고화질을 나타내는 새로운 image coder를 이 논문에서 논한다. 제안된 image coder는 constant segments를 가진 segmentation-based image coding technique의 문제들을 다음과 같은 방법론을 제안함으로써 해결하였다. Image를 HVS으로 보았을 때 degree of roughness에 관하여 textually homogeneous regions으로 segmentation하였다. Fractal dimension을 roughness of textural regions을 측정하기 위하여 사용하였다. Segmentation은 fractal dimension을 thresholding하여 textural regions이 three texture classes로 분류하였다(perceived constant intensity, smooth texture, and rough texture). High compression을 가지는 고질화의 image coder는 각각의 segment boundary와 각각의 texture class에 효율적인 coding technique를 적용 함으로 얻었다.

• 한국소성가공학회:학술대회논문집
• /
• 한국소성가공학회 2003년도 추계학술대회논문집
• /
• pp.34-38
• /
• 2003

In metal working, cold forging that has profit to satisfy dimension accuracy is using in various manufacturing products. Recently, most of the interest thing is precision forging of gear. Gear forging product is more strength than broaching gear, and it has many advantages with reduction of factory expenses. The reason of difficulty to improve accuracy of gear dimension compare to another products is the dimension accuracy is very high, approximately 10$\mu\textrm{m}$, and because die of involute teeth and elastic strain of forged tool differ from standard curve. This paper represent quantitative analysis of die and teeth of forged tool, namely difference of curves, with experiments and analyze the factor of dimension gap, finally, will design compensated involute curve.