• The Transactions of the Korea Information Processing Society
• /
• v.2 no.6
• /
• pp.937-946
• /
• 1995

We proposed and efficient block segmentation and classification method for the document analysis using SGLDM(spatial gray level dependence matrix) and BP (back Propagation) neural network. Seven texture features are extracted directly from the SGLDM of each gray-level block image, and by using the nonlinear classifier of neural network BP, we can classify document blocks into 9 categories. The proposed method classifies the equation block, the table block and the flow chart block, which are mostly composed of the characters, out of the blocks that are conventionally classified as non-character blocks. By applying Sobel operator on the gray-level document image beforebinarization, we can reduce the effect of the background noises, and by using the additional horizontal-vertical smoothing as well as the vertical-horizontal smoothing of images, we can obtain an effective block segmentation that does not lead to the segmentation into small pieces. The result of experiment shows that a document can be segmented and classified into the character blocks of large fonts, small fonts, the character recognigible candidates of tables, flow charts, equations, and the non-character blocks of photos, figures, and graphs.

• The Transactions of the Korea Information Processing Society
• /
• v.4 no.8
• /
• pp.2017-2024
• /
• 1997

The main focus of this paper is the recognition of printed Hangul documents in terms of typeface, character size and character slope for IICS(Intelligent Image Communication System). The fixed-size blocks extracted from documents are analyzed in frequency domain for the typeface classification. The vertical pixel counts and projection profile of bounding box are used for the character size classification and the character slope classification, respectively. The MLP with variable hidden nodes and error back-propagation algorithm is used as typeface classifier, and Mahalanobis distance is used to classify the character size and slope. The experimental results demonstrated the usefulness of proposed system with the mean rate of 95.19% in typeface classification. 97.34% in character size classification, and 89.09% in character slope classification.

• The Transactions of the Korea Information Processing Society
• /
• v.5 no.10
• /
• pp.2600-2608
• /
• 1998

This paper proposes an algorithm for off line recognition of handwritten characters, especially effective for large-set characters such as Korean and Chinese characters. The algorithm is based on a minimum distance dlassification method which is simple and easy to implement but suffers from low recognition performance. Two strategies have been developed to improve its performance; one is multi-stage pre-classification and the other is candicate reordering. Effectiveness of the algorithm has been proven by and experimet with the samples of 574 classes in a handwritten Korean character catabase named PE02, where 86.0% of recognition accuracy and 15 characters per second of processing speed have been obtained.

• Proceedings of the Korean Information Science Society Conference
• /
• 2000.10b
• /
• pp.389-391
• /
• 2000

본 논문을 자연여상에서 문자열의 위치를 찾아내는데 모폴로지 연산인 WTH(white top-hats)과 BTH(black top-hars)을 사용하였다. 기존의 자연영상에서의 문자열추출은 칼라양자화방법 경우 각 칼라공간에서 문자열 추출과정을 반복 적용하거나 모델기반방법의 경우 문자열의 획의 크기나 특징에 따라서 하나의 영상을 여러 개로 분리 적용하여 추가적인 계산비용을 필요로 한다는 점을 개선하고 공간적 변화도를 이용하여 영상을 직접 처리하는 경우 최소 문자열 후보영역을 찾기 위한 프로세스를 다시 적용해야 한다는 점을 개선하였다. 자연영상에 문자열의 위치를 대략적으로 찾아내기 위해 WTH+BTH을 적용하여 그 결과로 문자열의 대략적 위치와 최소문자열후보영역을 동시에 얻을 수 있다. 문자열이 가지는 특성을 적용하여 문자열-비문자열 분류과정을 적용하고 후처리를 통해 완전한 문자열의 위치를 보여준다.

• Cross-Cultural Studies
• /
• v.41
• /
• pp.163-175
• /
• 2015

In recent years, there has been significant interest and research on the Comprehensive Index of Character in the Warring States Period, which has resulted in many scholarly achievements. Hence, it is necessary to comprehensively categorize and summarize these achievements. Some researchers have already done a good job in this categorizing and summarizing. However, there is still work to be done in the comprehensive collation of the seal character research of the Warring States Period. The purpose of this essay is to regionalize the "SiMaZhiXi" Seal collected by "GuXiHuiBian (古璽彙編)" 0024. The number 0024 seal is an ancient seal collected by "Guxihuibian" (古璽彙編). In general, this is a Chu seal from the Warring States Period. But some scholars think that the seal is actually from the Qi or Yan States.

• Journal of KIISE
• /
• v.44 no.12
• /
• pp.1275-1281
• /
• 2017

The supervised learning approach is suitable for classification of insulting sentences, but pre-decided training sentences are necessary. Since a Character-level Convolution Neural Network is robust for each character, so is appropriate for classifying abusive sentences, however, has a drawback that demanding a lot of training sentences. In this paper, we propose transfer learning method that reusing the trained filters in the real classification process after the filters get the characteristics of offensive words by generated abusive/normal pair of sentences. We got higher performances of the classifier by decreasing the effects of data shortage and class imbalance. We executed experiments and evaluations for three datasets and got higher F1-score of character-level CNN classifier when applying transfer learning in all datasets.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2015.10a
• /
• pp.95-98
• /
• 2015

문자 인코딩은 컴퓨터에 저장하거나 네트워크상에서 전송하기 위해 문서를 이진화 하는 방법이다. 문자 인코딩은 고유의 문자 코드 테이블을 이용하여 문서를 이진화 하기 때문에, 문서에 적용된 문자 인코딩과 다른 문자 인코딩을 이용하여 디코딩 하면 원본과 다른 문서가 출력되어 문서를 읽을 수 없게 된다. 따라서 문서를 읽기 위해서는 문서에 적용된 문자 인코딩을 알아내야 한다. 본 논문에서는 문서의 문자 인코딩을 자동으로 판별하는 방법을 제시한다. 제안하는 방법은 이스케이프 문자를 이용한 판별법, 문서에 나타난 코드 값 범위 판별법, 문서에 나타난 코드 값의 특징 판별법, 단어 데이터베이스를 이용한 판별법과 같은 여러 단계를 걸쳐 문서에 적용된 문자 인코딩을 판별한다. 제안하는 방법은 문서를 언어별로 분류하여 문자 인코딩을 판별하기 때문에, 높은 문자 인코딩 인식률을 보인다.

• Smart Media Journal
• /
• v.7 no.4
• /
• pp.24-29
• /
• 2018

Text analysis technique for natural language processing in deep learning represents words in vector form through word embedding. In this paper, we propose a method of constructing a document vector and classifying it into spam and normal text message, using word embedding and deep learning method. Automatic spacing applied in the preprocessing process ensures that words with similar context are adjacently represented in vector space. Additionally, the intentional word formation errors with non-alphabetic or extraordinary characters are designed to avoid being blocked by spam message filter. Two embedding algorithms, CBOW and skip grams, are used to produce the sentence vector and the performance and the accuracy of deep learning based spam filter model are measured by comparing to those of SVM Light.

• Journal of Intelligence and Information Systems
• /
• v.10 no.3
• /
• pp.1-10
• /
• 2004

In this paper, a new character segmentation algorithm of machine printed character recognition is proposed. The new approach of the proposed character segmentation algorithm overcomes the weak points of both feature-based approaches and recognition-based approaches in character segmentation. This paper defines side profiles of touching characters. The character segmentation algorithm gives a candidate single character in touching characters by side profiles, without any help of character recognizer. It segments touching characters and decides the candidate single character by side profiles. This paper also defines cutting cost, which makes the proposed character segmentation find an optimal segmenting path. The performance of the proposed character segmentation algorithm in this paper has been obtained using a real envelope reader system, which can recognize addresses in U.S. mail pieces and sort the mail pieces. 3359 mail pieces were tested. The improvement was from $68.92\%\;to\;80.08\%$ by the proposed character segmentation.

• Proceedings of the Korean Information Science Society Conference
• /
• 2004.04b
• /
• pp.751-753
• /
• 2004

PDA에서 사용하는 대부분의 CPU는 실수 연산 구성요소(float computation component)가 없는 정수(integer) CPU를 사용한다. 인공 신경망(neural network)과 같은 실수 연산이 많은 알고리즘은 PDA에서 많은 수행시간을 가진다. 본 논문에서는 이런 단점을 해결하기 위해 무선 랜(LAN)으로 연결된 Client(PDA)/Server(PC) 구조를 이용한 효과적인 문자 추출 시스템을 제안한다. Client(PDA)는 대략적인 문자 추출 결과를 JPEG으로 압축하여 전송속도를 최소화한다. Server(PC)는 Client(PDA)의 결과를 바탕으로 정밀한 문자 영역 추출을 위해, 텍스춰 분류 방법과 연결 성분 분석 방법을 이용한다. 실험에서 제안한 방법은 속도뿐만 아니라 문자 추출에서도 효과적이었다.