• 지능정보연구
• /
• 제6권1호
• /
• pp.65-72
• /
• 2000

본 논문에서는 쿼리 영상에 대하여 가장 정확하게 정합되는 영상을 찾기 위한 새로운 이진 영상 정합 방법인 퍼지 기반 하우스도르프 방법을 제안한다, 먼저 하우스도르프 거리를 이용하여 최소거리 분포를 얻은 후 반경에 해당하는 집함의 개수를 이용하여 소속함수로 표현한다. 제안한 방법에서는 소속함수로 정의된 거리 분포에 대하여 퍼지 추론과정을 도입하여 최종적인 정합 후보를 구하게 된다. 제안된 방법을 실제 잡음이 부가된 얼굴 영상과 문자 인식에 적용하여 그 성능을 검증하였다.

• 한국통신학회논문지
• /
• 제22권4호
• /
• pp.726-736
• /
• 1997

The textual image compression by pattern matching is a coding scheme that exploits the correlations between patterns. When we compress the Hangul (Korean character) text by patern matching, the collerations between patterns may decrease due to randoem contacts between phonemes. Therefore in this paper we separate connected phonemes to exploit effectively the corrlation between patterns by inducting the amtch. In the process of sequation, we decide whether the patterns have vowel component or not, and then vowels connected with consonant ae separated. When we compare the proposed algorithm with the existing algorith, the compression ratio is increased by 1.3%-3.0% than PMS[5] in lossy mode, by 3.4%-9.1% in lossless mode than that of SPM[7] which is submitted to standard committe for second generation binary compression algorithm.

• 반도체디스플레이기술학회지
• /
• 제19권4호
• /
• pp.130-134
• /
• 2020

This paper proposes a new method for the digit recognition on a 7-segment display. The proposed method uses morphological processing that dilates segments of digits and connects them into strokes. The digits are extracted by connected component analysis and finally, template matching method recognizes the extracted digits. The proposed method is implemented using C language in Raspberry Pi 4 system with a camera module for a real-time image processing. Experiments were conducted by using various 7-segment LED displays and 7-segment mono LCD displays. The results show that the proposed method is successful for the digit recognition on the 7-segment displays.

• 대한전자공학회논문지SP
• /
• 제41권6호
• /
• pp.9-18
• /
• 2004

본 논문에서 우리는 시차공간영상 (disparity space image)을 이용한 새로운 스테레오 정합 방법을 제안한다. 본 논문에서는 실측 제어점 대신 영상 안에서 특징점을 추출하여 정합영역을 설정하였고, 간단한 비용행렬을 정의하였다. 그리고 정합 속도를 증가시키기 위해 화소 단위의 정합을 이용하였다. 화소 단위의 정합은 정합 속도는 향상시키지만, 주변 영역을 이용하지 않기 때문에 정합의 정확성이 낮아진다. 이러한 단점을 보완하기 위해 시차공간영상의 특성을 이용하여 정합경로를 확대시켜서 주변 영역을 최대한 이용하였다. 또한, 현재 위치에서 이전과 이후의 시차공간영상을 체적화하여 전후의 시차공간영상을 이용함으로써 정합의 정확도를 더욱 향상시키는 정합 보정 모듈을 추가하였다. 본 논문에서 제안된 알고리즘은 다양한 영상에 적용하여 실험하였고, 그 결과 제안된 방법의 우수한 성능을 확인할 수 있었다.

• 한국정보과학회논문지:컴퓨팅의 실제 및 레터
• /
• 제16권5호
• /
• pp.546-550
• /
• 2010

본 논문에서는 모바일 폰에서 오프라인 필기체 과분할 인식의 후처리 방법에 관하여 논하였다. 제안된 방법은 조합 행렬 생성, 문자 조합 필터링, 문자 유사도 측정으로 구성된다. 조합 행렬 생성 과정은 각각의 조각의 인식 결과로부터 생성가능한 모든 조합 행렬을 계산하는 부분이며 조합 행렬을 그래프로 구성하게 된다. 문자 조합 필터링 과정은 그래프의 노드들과 단어 사전을 비교하여 불필요한 노드를 삭제하는 과정이며 문자 유사도 측정과정은 단어 사전의 각각의 단어들과 Levenshtein 거리(distance)를 계산하여 최적의 후처리 결과를 추출하게 된다. 제안된 방법의 인식률은 85.8%의 정확도를 보였다.

• International Journal of Contents
• /
• 제17권4호
• /
• pp.1-15
• /
• 2021

Named entity recognition (NER) is a crucial task for NLP, which aims to extract information from texts. To build NER systems, deep learning (DL) models are learned with dictionary features by mapping each word in the dataset to dictionary features and generating a unique index. However, this technique might generate noisy labels, which pose significant challenges for the NER task. In this paper, we proposed DL-dictionary features, and evaluated them on two datasets, including the OntoNotes 5.0 dataset and our new infectious disease outbreak dataset named GFID. We used (1) a Bidirectional Long Short-Term Memory (BiLSTM) character and (2) pre-trained embedding to concatenate with (3) our proposed features, named the Convolutional Neural Network (CNN), BiLSTM, and self-attention dictionaries, respectively. The combined features (1-3) were fed through BiLSTM - Conditional Random Field (CRF) to predict named entity classes as outputs. We compared these outputs with other predictions of the BiLSTM character, pre-trained embedding, and dictionary features from previous research, which used the exact matching and partial matching dictionary technique. The findings showed that the model employing our dictionary features outperformed other models that used existing dictionary features. We also computed the F1 score with the GFID dataset to apply this technique to extract medical or healthcare information.

• ETRI Journal
• /
• 제31권5호
• /
• pp.545-553
• /
• 2009

String matching is a fundamental element of an important category of modern packet processing applications which involve scanning the content flowing through a network for thousands of strings at the line rate. To keep pace with high network speeds, specialized hardware-based solutions are needed which should be efficient enough to maintain scalability in terms of speed and the number of strings. In this paper, a novel architecture based upon a recently proposed data structure called the Bloomier filter is proposed which can successfully support scalability. The Bloomier filter is a compact data structure for encoding arbitrary functions, and it supports approximate evaluation queries. By eliminating the Bloomier filter's false positives in a space efficient way, a simple yet powerful exact string matching architecture is proposed that can handle several thousand strings at high rates and is amenable to on-chip realization. The proposed scheme is implemented in reconfigurable hardware and we compare it with existing solutions. The results show that the proposed approach achieves better performance compared to other existing architectures measured in terms of throughput per logic cells per character as a metric.

• 제어로봇시스템학회논문지
• /
• 제16권9호
• /
• pp.913-917
• /
• 2010

To inspect the defects of printed markings on the surface of IC package, the OCV (Optical Character Verification) method based on NCC (Normalized Correlation Coefficient) pattern matching is widely used. In order to detect the micro pattern defects appearing on the small portion of the markings, a Partitioned NCC pattern matching method was proposed to overcome the limitation of the NCC pattern matching. In this method, the reference pattern is first partitioned into several blocks and the NCC values are computed and are combined in these small partitioned blocks, rather than just using the NCC value for the whole reference pattern. In this paper, we proposed a method to decide the proper number of partition blocks and a method to inspect and combine the NCC values of each partitioned block to identify the defective markings.

• 조명전기설비학회논문지
• /
• 제24권12호
• /
• pp.38-46
• /
• 2010

For the development of 100[W] stage lighting, quantitative and uniform color mixing that applied through color adopted Simple-Inverse matching so that color mixing can be possible along Black Body Locus. R,G,B(Red, Green, Blue) LED(Light Emitting Diode) arrangement through LED package character analysis, LED module, and the characteristic of device were considered for uniform color mixing. A distance changeable optical device was built to assure high uniformity and high diffusion of not only the middle of diffusion side but also the border side. Also, we developed the control power circuit that can expand up to 6 channels which are possible for quantitative color mixing, and the high uniformity and high quantified border light for color mixing control and the verification of color mixing characteristics by composing GUI(Graphical user interface) including color mixing simulator. By presenting the experimental results of light color control, we proved the usefulness of our developed border light and the proposed color mixing method.

• 정보처리학회논문지B
• /
• 제12B권3호
• /
• pp.357-364
• /
• 2005

본 논문에서는 두 단계 이미지 매칭을 이용하여 한글 문서영상에서 사용자 검색어를 빠르고 정확하게 검색할 수 있는 시스템을 제안한다. 본 시스템은 문자 분리, 검색어 영상 생성, 특징 추출 그리고 이미지 매칭 과정으로 구성된다. 매칭 과정에서 차원이 다른 두 가지 특징 벡터를 이용한다. 8쪽 분량의 문서 영상을 한국정보과학회 웹사이트에서 다운로드하였고, 그 문서로부터 1600개의 한글단어 영상을 획득하여 실험데이터로 사용하였다 그 결과 제안한 시스템은 기존에 제안된 영상-기반 한글 단어 검색 시스템보다 성능이 크게 향상되었음을 알 수 있었다.