• 한국산학기술학회논문지
• /
• 제20권9호
• /
• pp.49-55
• /
• 2019

본 논문에서는 효과적인 마커기반의 증강현실 구현을 위하여 영상 내 객체의 분포에 대한 분석과 반복 패턴을 포함하는 영상의 분류를 통한 마커영상의 평가 방법을 제안한다. 객체의 분포는 영상의 부분적 가림 현상에 따라 객체추적성능에 영향을 미치기 때문에 특징점 좌표의 분산을 이용하여 가림 현상에 취약한 마커영상을 구분할 수 있도록 하였고, 일반 영상과 반복 패턴을 포함하는 영상의 특징점 기술자 벡터의 분포가 현저하게 다르다는 사실에 기반하여 객체의 인식 및 추적에 적합한 영상을 구분할 수 있는 방법을 제안한다. 다양한 실험 결과 제안하는 마커 평가 방법이 가림 현상에 취약한 영상 및 반복 패턴 영상을 구분하는데 우수한 성능을 보이는 것을 확인하였다. 또한 마커영상에 대한 객체 추적 등의 안정성 측면에서 SURF보다 SIFT 기법이 우수한 성능을 보임을 확인할 수 있었다. 이러한 결과를 이용하여 다양한 종류의 마커영상에 대한 적합성 정보를 사용자에게 제공함으로써 효과적인 증강현실 시스템을 구현할 수 있을 것으로 판단된다.

• 한국정보통신학회논문지
• /
• 제10권5호
• /
• pp.871-876
• /
• 2006

본 논문은 공간 부호화 패턴들을 이용하여 3차원 얼굴 정보를 정확하게 측정하기 위하여 초기 얼굴 패턴 영상으로부터 이미지 패턴을 검출하기 위한 새로운 알고리즘을 제안한다. 획득된 영상이 불균일하거나 패턴의 경계가 명확하지 않으면 패턴을 분할하기가 어렵다. 그리고 누적된 오류로 인하여 코드화가 되지 않는 영역이 발생한다. 본 논문에서는 이러한 요인에 강하고 코드화가 잘 될 수 있도록 FCM 클러스터링 방법을 이용하였다. 패턴 분할을 위하여 클러스터는 2개, 최대 반복횟수는 100, 임계값은 0.00001로 설정하여 실험하였다. 제안된 패턴 분할 방법은 기존 방법들(Otsu, uniform error, standard deviation, Rioter and Calvard, minimum error, Lloyd)에 비해 8-20%의 분할 효율을 향상시켰다.

• 대한의용생체공학회:의공학회지
• /
• 제10권2호
• /
• pp.125-130
• /
• 1989

In the present study, we described a test to explore the function of optokinetic system which subjected to a full visual field stimulation using two different stimulus images patterns. Our results were interesting in a point of view that the stimulation image pattern had non- neglisible influence on the optokinetic response and that in a bidimensionnel image such as the randomly distributed spots images pattern the linearity of system was assured upto the stimulus velocity of about 50 deg/sec for normal subject. As for measuring human optokinetic after nystagmus, the regular stripes pattern was rather desirable than the randomly distributed spots pattern in this study.

• 제어로봇시스템학회논문지
• /
• 제15권5호
• /
• pp.548-554
• /
• 2009

We propose an automatic inspection system for hologram with multiple patterns. The system hardware consists of illuminations, camera, and vision processor. Multiple illuminations using LEDs are arranged in different directions to acquire each image of patterns. The system software consists of pre-processing, pattern generation, and pattern matching. The acquired images of input hologram are compared with their reference patterns by developed matching algorithm. To compensate for the positioning error of input hologram, reference patterns of hologram for different position should be generated in on-line. We apply a frequency transformation based CGH(computer-generated hologram) method to generate reference images. For the fast pattern matching, we also apply the matching method in the frequency domain. Experimental results for hologram of Korean currency are then presented to verify the usefulness of proposed system.

• 대한전기학회논문지:시스템및제어부문D
• /
• 제50권10호
• /
• pp.473-478
• /
• 2001

Image processings using the two-dimensional wavelet transform (2DWT) have been a very active research area in recent years because the 2DWT possess many good properties. However, the discrete 2DWT can not be used for pattern recognition directly because it does not have the translation property. In this paper, we show why conventional discrete two-dimensional wavelet transforms cannot be used for pattern recognitions directly. Then, we propose a new method that makes it possible to use discrete 2DWT to pattern recognition without modification of standard pyramidal algorithms. The main idea of our method is to postprocess the wavelet transformed images using the second-order neural network. To justify the validity of the method, evaluations with test images were performed. The effectiveness of the method can be shown by the evaluation results.

• 융합신호처리학회논문지
• /
• 제8권3호
• /
• pp.185-191
• /
• 2007

윈격탐사 영상은 파장대에 따라 나누어진 여러 개의 밴드로부터 수집된 다중분광 이미지 데이터이다. 위성영상 분류는 원격탐사 처리 과정에 있어서 가장 중요한 분석 기법으로써 영상을 구성하는 각각의 화소들 중 비슷한 분광 특성을 갖는 것끼리 집단화시켜주는 방법이다. 본 논문에서는 PFCM 알고리즘을 응용한 원격탐사 영상의 패턴분류 방법에 관하여 연구하였다. PFCM 알고리즘은 각 데이터와 특정 클러스터 중심과의 거리에 대한 소속정도를 고려한 FCM 클러스터링 알고리즘과 데이터와 해당 클러스터 중심과의 거리에 의존하여 패턴의 전형성(typicality)을 고려한 PCM 클러스터링 알고리즘을 결합한 방법이다. 본 연구에서는 분류 항목별 학습데이터를 선정한 후 이를 PFCM 알고리즘에 적용하여 감독분류를 수행하였다. Landsat TM과 IKONOS 원격탐사 위성영상을 이용하여 PFCM 알고리즘의 적용성을 검증하였다. PFCM 알고리즘을 이용한 감독분류는 PCM, FCM 분류방법보다 좋은 결과를 보여주었으며, 또한 전통적인 분류방법인 최대우도분류보다도 정확도가 더 높은 결과를 보여주었다.

• 한국생활과학회지
• /
• 제10권4호
• /
• pp.399-409
• /
• 2001

The purpose of this study was to evaluate the preference of image for modern application of Korean traditional patterns. A survey was conducted using the random selection among female undergraduate students in Daegu city. The degree of interest and preference in Korean traditional style or something like that measured by 5 scale method. And then they were classified into two groups which were interest/non-interest group, and preference/non-preference group. The image of Korean traditional patterns consisted of semantic differential scales. Frequency, percentage and mean were analyzed, for difference of groups t-test was analyzed. The results were as fellows; 1. For the degree of interest for Korean traditional patterns, it was showed that 53.8% of total respondents took interest and about 40.4% of them had preference for traditional patterns. the correlation coefficient of the degree of interest and preference was 0.782(p<0.01) and showed that the positive correlation was high. 2. Among 20 kinds of Korean traditional patterns, the degree of preference for the patterns of plants and nature was quite high whereas that for the patterns of geometrical things was low relatively. 3. It was evaluated that pattern of nature was fresh, refined and womanly image generally. It was evaluated that pattern of plants was womanly, fresh, weak, light and soft image and that of animals was heavy, splendid, high-class, manly, strong and positive image. It was evaluated that pattern of geometrical things was the most refined image and high-class, rigid and strong. 4. The statistical significance of mean between interest/non-interest group was showed statistically in the patterns of clouds, mountains, lotus, apricot, orchid, dragon, phoenix and bogey. In case of pattern of orchids, the degree of preference was most different between interest/non-interest group. 5. The pattern of plants showed the most different evaluation for images between interest/non-interest group. For refined/old-fashioned polar adjective images, the interest group evaluated the pattern of plants more refined. 6. For pattern of orchids, the difference of degree of preference between preference/non-preference group was most remarkable in Korean traditional patterns. 7. The pattern of geometrical things showed the most different evaluation for images between preference/non-preference group. For warm/cool polar adjective images, the preference group evaluated the pattern of geometrical things cooler.

• 대한전자공학회논문지SP
• /
• 제42권6호
• /
• pp.133-140
• /
• 2005

본 논문에서는 3차원 얼굴 인식기에서의 입력 데이터로 적합한 얼굴 데이터의 유클리디안 복원 시스템을 제안한다. 카메라 영상을 통한 유클리디안 복원을 위해서는 카메라의 보정정보와 복원할 특징 점의 대응 쌍 정보가 필요한데, 보정정보는 시스템의 변경이 없다면 불변한다. 따라서 고속의 유클리디안 복원을 위해서는 스테레오 영상간의 대응 쌍 정보 획득이 가장 중요하다. 시스템은 두 개의 카메라와 프로젝터 한 대로 구성하며, 피사체에 패턴을 투사하고 두 개의 카메라로부터 영상을 획득함으로써 동작한다. 사용 패턴은 단일 투사 줄무늬 패턴을 사용하며 줄무늬의 색도 정보와 채도 정보를 이용하여 '절대 코드 패턴'을 생성한다. 사용 시스템은 두 개의 카메라를 사용하기 때문에 색도 정보와 채도 정보가 비슷한 두 개의 영상을 얻을 수 있으며 이것은 획득한 두 개의 영상에 있는 줄무늬들에 대해 동일한 절대 라벨을 얻을 수 있게 한다. 절대 라벨링 된 줄무늬는 에피폴라 라인(epipolar line)상의 점들과 일대일 정합을 통해 고속의 3차원 데이터 복원을 가능하게 한다. 제안 방식은 얼굴 데이터에 적용되었으며 정확도와 총 소요 시간의 측정을 통해 제안 방식이 기존 방식에 비해 정확도 면에서 뒤지지 않으면서도 복원 속도 면에서 장점을 가짐을 확인할 수 있었다.

• 한국정보시스템학회지:정보시스템연구
• /
• 제27권3호
• /
• pp.123-140
• /
• 2018

Purpose The purpose of this study is to explore the optimal trading frequency which is useful for stock price prediction by using deep learning for charting image data. We also want to identify the appropriate time for accurate forecasting of stock price when performing pattern analysis. Design/methodology/approach In order to find the optimal trading frequency patterns and forecast timings, this study is performed as follows. First, stock price data is collected using OpenAPI provided by Daishin Securities, and candle chart images are created by data frequency and forecasting time. Second, the patterns are generated by the charting images and the learning is performed using the CNN. Finally, we find the optimal trading frequency patterns and forecasting timings. Findings According to the experiment results, this study confirmed that when the 10 minute frequency data is judged to be a decline pattern at previous 1 tick, the accuracy of predicting the market frequency pattern at which the market decreasing is 76%, which is determined by the optimal frequency pattern. In addition, we confirmed that forecasting of the sales frequency pattern at previous 1 tick shows higher accuracy than previous 2 tick and 3 tick.

• 한국의류산업학회지
• /
• 제17권4호
• /
• pp.524-540
• /
• 2015

This study develops textile pattern design ideas created through the visualization of music. Methods of auditory and synesthesia were employed to analyze various attributes of popular music genres and appoint language image, shape image, and color image to obtain their interrelationships. This study provides data that can be used to express emotional images on textile print pattern designs. This research used different genres of popular music as stimuli. The language image was extracted and introduced to the overall color scheme; in addition, the color image was verified. The analysis of the color image was executed by applying it with the color set image scale of I.R.I colors. Then, the color image of the target genre of popular music was examined and analyzed through a color tone system. The preference in shape image was realized through visual images based on basic principles of points, lines, and sides composition; subsequently, an analysis of the emotional image of popular music followed. An examination of the emotional images of different popular music genres have led to the discovery that language image, color image, and shape image all share a common emotional image. There was also a realization that similarity and interrelationship exists in language, color, and shape images experienced by listening to popular music.