• 전자공학회논문지B
• /
• 제32B권3호
• /
• pp.522-532
• /
• 1995

Multilayer neural networks with backpropagation learning algorithm are widely used for pattern classification problems. For many real applications, it is more important to reduce the misclassification rate than to increase the rate of successful classification. But multilayer perceptrons(MLP's) have drawbacks of slow learning speed and false convergence to local minima. In this paper, we propose a new method for character recognition problems with a single-layer network and double rejection mechanisms, which guarantees a very low misclassification rate. Comparing to the MLP's, it yields fast learning and requires a simple hardware architecture. We also introduce a new coding scheme to reduce the misclassification rate. We have prepared two databases: one with 135,000 digit patterns and the other with 117,000 letter patterns, and have applied the proposed method for printed character recognition, which shows that the method reduces the misclassification rate significantly without sacrificing the correct recognition rate.

• 산업공학
• /
• 제13권2호
• /
• pp.188-194
• /
• 2000

A pattern recognition system is proposed to recognize characters in any type of Giro. The system consist of the character segmentation and the character recognition. Positional features from two round markers at the upper-right part and lower-left part of Giro is used for extracting character strings from images and RLE analysis is used if there are no round markers. A multi step combined method, which use a structural method and a statistical method, is used to improve recognition. The structural method apply rules on each characters, whereas a statistical method gives a different weighting vector to each pixel for improving the classification performance in regard to noises and distortions. The experimental results show that the proposed combined method has higher recognition rate, over than 98% even in cases that images are rotated about 10 degrees as well as have noises.

• 전자통신동향분석
• /
• 제37권5호
• /
• pp.22-32
• /
• 2022

Optical character recognition is a primary technology required in different fields, including digitizing archival documents, industrial automation, automatic driving, video analytics, medicine, and financial institution, among others. It was created in 1928 using pattern matching, but with the advent of artificial intelligence, it has since evolved into a high-performance character recognition technology. Recently, methods for detecting curved text and characters existing in a complicated background are being studied. Additionally, deep learning models are being developed in a way to recognize texts in various orientations and resolutions, perspective distortion, illumination reflection and partially occluded text, complex font characters, and special characters and artistic text among others. This report reviews the recent deep learning-based text detection and recognition methods and their various applications.

• 복식
• /
• 제59권6호
• /
• pp.29-40
• /
• 2009

On April 28th of 2006, a set of Buddha's reliquary was excavated from the pagoda in n Nacsan Temple. According to the record, the relics were put in the pagoda in the 18th year of King SookJong in the Chosun Dynasty (1692). The present paper examines ten pieces of wrapping clothes covering reliquary found in the pagoda. They are especially precious data in the history of textiles because they were blocked off from outside and was preserved in good condition with vivid colors still remaining after more than 300 years. Of the ten pieces of wrapping cloth, five were double-layered and the other five were single-layered. They include 15 pieces of silk fabric but, excluding repeated use of the same silk fabric, the total of 11 pieces of silk fabric were examined. All 11 kinds of silk fabric were patterned, 9 of which were Satin and the other 2 were Twill. Of the 9 Satin pieces, 8 pieces were 5-end satin which had the ground of 5-end warp satin with the figure of 5-end weft satin. The remaining 1 Satin piece were more splendid with prominent figures by using warp and weft of different colors. The 2 Twill pieces used twill weave-the ground was 3-end warp twill and the figures were 5-end weft twill. Both of the Twill pieces were weaved with character patterns, partly using wrapped gold thread as supplementary weft. The patterns of 11 pieces of silk fabric include flower, dragon/phoenix, cloud, and geometric patterns. Five were flower patterns, three were dragon/phoenix patterns, two were geometric pattern, and one was cloud pattern. In addition, various treasure patterns, character patterns were utilized as supplementary patterns. The flower and phoenix patterns reflect characteristics of the textiles of the 17th century whereas check pattern and cloud pattern were very unique.

• 한국응용곤충학회지
• /
• 제37권2호
• /
• pp.163-170
• /
• 1998

파밤나방(Spodoptera exigua (H bner)) 유충의 형태적 유전지표를 규명하기 위해 유충의 체색과 줄무늬 변이에 관하여 분석하였다. 유충 체색은 먹이 종류에 따라 다양했다. 유충 줄무늬는 배선과 측선의 존재에 따라 3종류의 형질 개체로 나뉘었다. 3줄무늬(배선과 측선 모두존재), 1줄무늬(배선만존재), 0줄(배선과 측선 모두 없음). 이들 형질이 유전적 영향에 있는지를 조사하기 위해 3줄과 1줄집단으로 집단선발한 결과 각 선발 형질의 비율이 증가했다. 3줄과 1줄집단을 상호교배하였을 때 3줄 형질이 우성을 보였다. 줄무늬 형질에 있어서 협의의 유전력(h$^2$)은 $0.50\pm$0.42로 산출되었다. 0줄집단에서 암컷수가 수컷수에 비해 약2배 많았다. 환경적 요인을 조사하기 위해 동일 집단을 3종류의 먹이에서 사육했을때 인공사료와 상치로 키운 집단간에는 유충줄무늬에는 유의성있는 차이는 없지만 파로 사육된 집단에서는 3줄개체의 비율이 낮아져 차이를 보였다. 이들 유충의 줄무늬는 유충과 용의 발육속도 및 내한성과 연관성을 보였으나 살충제 감수성과는 무관함을 나타냈다.

• 전기학회논문지
• /
• 제57권2호
• /
• pp.320-326
• /
• 2008

In this paper, we propose a new algorithm emulating the DNA characteristics for noise-tolerant pattern matching problem on digital system. The digital pattern matching becomes core technology in various fields, such as, robot vision, remote sensing, character recognition, and medical diagnosis in particular. As the properties of natural DNA strands allow hybridization with a certain portion of incompatible base pairs, DNA-inspired data structure and computation technique can be adopted to bio-signal pattern classification problems which often contain imprecise data patterns. The key feature of noise-tolerance of DNA computing comes from control of reaction temperature. Our hardware system mimics such property to diagnose cardiovascular disease and results superior classification performance over existing supervised learning pattern matching algorithms. The hardware design employing parallel architecture is also very efficient in time and area.

• 한국통신학회논문지
• /
• 제38A권12호
• /
• pp.983-990
• /
• 2013

본 논문에서는 RFID tag의 한 종류이며 가격경쟁력이 뛰어난 chipless RFID tag를 한글의 형태로 제안하고 있다. 한글형 chipless RFID tag는 육안으로도 인식이 가능하면서 동시에 기계적 언어로도 인식이 가능하다는 장점을 가지며, 이러한 한글형 chipless RFID tag를 위해 대표적인 한글 자음 10가지와 모음 10가지를 조합하여 34개의 문자를 시뮬레이션 하여 RCS값을 데이터 베이스화하였다. 시뮬레이션된 특정 문자의 RCS값에 잡음을 추가하여 실측 환경을 모델링하고 데이터베이스화 된 각 문자의 RCS 데이터와의 차에 대한 분산을 구하는 알고리즘을 통해 대조함으로써 해당 문자를 인식하는 방법을 사용하였다.

• Journal of Multimedia Information System
• /
• 제6권4호
• /
• pp.225-234
• /
• 2019

Deep learning has become increasingly popular in both academic and industrial areas nowadays. Various domains including pattern recognition, Computer vision have witnessed the great power of deep neural networks. However, current studies on deep learning mainly focus on quality data sets with balanced class labels, while training on bad and imbalanced data set have been providing great challenges for classification tasks. We propose in this paper a method of data analysis-based data reduction techniques for selecting good and diversity data samples from a large dataset for a deep learning model. Furthermore, data sampling techniques could be applied to decrease the large size of raw data by retrieving its useful knowledge as representatives. Therefore, instead of dealing with large size of raw data, we can use some data reduction techniques to sample data without losing important information. We group PCB characters in classes and train deep learning on the ResNet56 v2 and SENet model in order to improve the classification performance of optical character recognition (OCR) character classifier.

• 대한가정학회지
• /
• 제38권11호
• /
• pp.137-149
• /
• 2000

The Purpose of this study was to investigate actual condition about the modem adaptation of interior elements(floor, wall, ceiling, door & window) in traditional thatched roof house. The examined objects were interior space of 36 restaurants and cafes in Seoul and Kyung-Ki Do area. 1. Floor: Jang-pan was mostly alternated with linoleum which huts Jang-pan pattem. Wumul-maru was adapted from the original and Jang-maru was alternated with wood or linolium which has western state Jang-maru pattern. Mud was adapted from the original or alternated with slate stone or rough finish cement. 2. Wall: Rice proper was alternated with rice paper book witch has chinese character, paper for parcels or modem wall paper. Plaster-white paint or white handy coat. Mud-mud color paint or bamboo stick witch located in the mud wall orginal. Log-half cut log. Wooden board-without cross bar or irregular form. 3. Ceiling: Yondung-Chongang was mostly adapted from the original and Banja-Chonjang was alternated with rice paper book which has Chinese character or modem wall paper. 4. Door and Window: Ttisal-mun and Panjang-mun were adapted from the original. Wan and A’character door and window were simplified character itself.

• 한국융합학회논문지
• /
• 제9권11호
• /
• pp.61-68
• /
• 2018

본 논문에서는 일상에서 획득할 수 있는 자연 영상에서 문자를 검출하기 위해 영상 개선 및 문자의 형태를 분석하여 문자를 검출하는 방법을 제안한다. 제안하는 방법은 자연 영상에서 문자로 인식될 영역의 검출률을 향상시키기 위해 객체부분의 경계를 언샤프 마스크를 사용하여 강조하였다. 향상된 객체의 경계 부분을 이용하여 영상의 문자 후보영역을 MSER(Maximally Stable Extermal Regions)을 이용하여 검출하였다. 검출된 문자 후보영역에서 실제 문자로 판단될 영역을 검출하기 위해 각 영역들의 형태를 분석하여 글자의 특성을 갖는 영역외의 비 문자영역을 제거하여 실제 문자영역 검출률을 높였다. 본 논문의 정량적 평가를 위해 문자 영역의 검출률과 정확도를 이용하여 기존의 방법들과 비교하였다. 실험결과 기존의 문자 검출 방법보다 제안하는 방법이 비교적 높은 문자영역의 검출률 및 정확도를 보였다.