• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.9
• /
• pp.84-93
• /
• 1993

Recently, several nonlinear shape normalization methods have been proposed in order to compensate for the shape distortions in handwritten characters. In this paper, we review these nonlinear shape normalization methods from the two points of view : feature projection and feature density equalization. The former makes feature projection histogram by projecting a certain feature at each point of input image into horizontal-or vertical-axis and the latter equalizes the feature densities of input image by re-sampling the feature projection histogram. A systematic comparison of these methods has been made based on the following criteria: recognition rate, processing speed, computational complexity and measure of variation. Then, we present the result of quantitative evaluation of each method based on these criteria for a large variety of handwritten Hangul syllables.

• Proceedings of the IEEK Conference
• /
• 2000.06c
• /
• pp.19-22
• /
• 2000

This paper presents new on-line handwritten algorithm for continuous alphabet uppercase characters. The algorithm is based on the idea that alphabet uppercase character consists of at most 4 strokes. It tries to determine the maximum output for a recognition result among outputs of four recognizers which have the capacity to discriminate the character using from 1 through 4 stroke information. The recognition module has 4 neural network based recognizers, which can recognize from 1 through 4 stroke character. We also use specialized post-processing techniques for improving the recognition performance. Trained on 440 input data and choosing 390 uppercase words for a recognition test we reached a 92% recognition rate.

• Proceedings of the KIEE Conference
• /
• 1991.07a
• /
• pp.900-904
• /
• 1991

In this paper, our study for the recognition of Hand-written Korean characters, Arabic numbers and alphabets by neural netwoks. This System extracts feature of character by using the MESH feature point of handwritten character, Arabic numbers and alphabets. To reduce the input image data, features are extracted from each input images. A MLP(multi-layer perceptron) with one hidden layer was trained with a modified BEP(back error propagation) algorithm. This method extracts feature sets of the characters directly from the scanner and can enhance computation speed without using the special preprocesses such as size normalization, smoothing, and thinning.

• Journal of the Korean Graphic Arts Communication Society
• /
• v.6 no.1
• /
• pp.41-57
• /
• 1988

Recently, it Is required to develop OCR(Optical Character Reader) along with the progress of the information processing system for Hangeul. Characters have to be recognized clearly so that OCR can be applied, Structure analysis method and lump method are used for the recognition of characters, and OCR is now available for the recognition of printed characters and handwritten alphanumeric characters having simple structure by them However, It is known that there should be much more study on the development of handwritten Hangout's OCR. This paper proposed a new method for the handwritten Hangout character recognition. The units of Initial consonant of Hangout are separated and then recognized from the utilization of the position- Information of Hangeul's units from the normalized patterns using the regression line theory. It is carried out for the extraction of the block which exists in the virtual Initial consonant region from the normalized input patterns and the calculation on maximum value (${\beta}$) of likelihood after comparing the features of separated subpattern with the initial consonant dictionary.

• Proceedings of the Korea Information Processing Society Conference
• /
• 2002.11c
• /
• pp.2285-2288
• /
• 2002

PDAs와 같은 휴대용 컴퓨터의 발전은 펜을 이용한 데이터 입력과 처리 기술에 대한 요구를 증가시키고 있다. 하지만 대부분의 PDAs용 애플리케이션들은 키보드와 마우스를 주된 입력 도구로 사용하는 데스크 탑 컴퓨터의 GUI와 입력 방식을 그대로 사용하고 있다. 본 논문에서는 펜 기반의 휴대용 컴퓨터에서 필기체 데이터 문서를 편집할 수 있도록 해주는 펜 제스쳐(pen gesture)기능 허용 텍스트 에디터를 개발하였다. 삽입, 삭제, 앞뒤 낱말 바꾸기, 줄 바꾸기 등의 제스쳐 인식 방법과 제스쳐 처리를 위해 스트록들(strokes)을 GCs(Gesture Components) 단위로 그룹화하기 위한 방법들을 제안하였다. 제스쳐 허용 에디터의 개발은 펜 기능만으로 문서 편집을 위한 모든 작업을 가능하게 해 줌으로 사용자의 편의를 극대화시켜 준다.

• 제어로봇시스템학회:학술대회논문집
• /
• 1996.10b
• /
• pp.466-469
• /
• 1996

In this paper, we present the global minimization condition by an informal analysis of the Langevine competitive learning neural network. From the viewpoint of the stochastic process, it is important that competitive learning guarantees an optimal solution for pattern recognition. By analysis of the Fokker-Plank equation for the proposed neural network, we show that if an energy function has a special pseudo-convexity, Langevine competitive learning can find the global minima. Experimental results for pattern recognition of handwritten numeral data indicate the superiority of the proposed algorithm.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.32B no.1
• /
• pp.147-153
• /
• 1995

In this paper, we propose the position- and size-independent handwritten on-line Korean character recognition system using multilayer neural networks which are trained with error back-propagation learning algorithm and the features of Hanguel consonants and vowels. Starting point, end point, and three vectors from starting point to end point of each stroke of characters inputted from mouse or tablet are applied as inputs of neural networks. If double consonants and vowels are separated by single consonants and vowels, all consonants and vowels have at most four strokes. Therefore, four neural networks learn the consonants and the vowels having each number of strokes. Also, we propose the algorithm of separating the consonants and vowels and constructing a character.

• Journal of the Korean Institute of Telematics and Electronics C
• /
• v.34C no.7
• /
• pp.61-69
• /
• 1997

With the vigorous researches in the character recognition, the need to recognize run-on multilingual handwritten characters is increasing to provide uses with more comfortable PUI(pen user interface) environments. In general, many intermediate word candidates word candidates are generated in run-on multilingual recognition because there is no information of ending position and linguistic kind of character. To remove unnecessary word candidates which are generated in run-on multilingual recognition, we classify them into two groups and select the best candidate among the word candidates in the group where the final characater is completed using 5 attributes. In this research, we propose a method in order to select the best one candidate. It is called WRM (Weighted ranking method). The weights are adaptively trained by LMS(Least mean square) learning rule. Results show that the abilities of decision makin gusing weights are much better than those not using weights.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.30B no.3
• /
• pp.50-59
• /
• 1993

There is basic problem with the current evaluation method for character recognition systems. The current method evaluates the average recognition rate by applying the test data to the target system. The average recognition rate tells no more than and no less than the overall performance and it depends on the data. In this paper we propose a testing method which will analyze the target system and point out its strong points and weak points. This can be made possible through using the data which are generated cy distorting the standard character images according to a carefully controlled manner. This paper will describe how to automatically generate such distorted images. Also we will show the method is actually effective and useful by applying it to evaluating existing recognition algorithms.

• Proceedings of the IEEK Conference
• /
• 1999.11a
• /
• pp.456-459
• /
• 1999

In this paper, we propose a method of unconstrained handwritten numeral recognition using image dithering and multiple modular MLPs. The set of sample numeral patterns is subdivided into clusters which are extended by their radius. On each extended cluster, we constructed MLPs network as the expert recognizer of corresponding cluster. The gating network is also trained by an MLPs to weigh the outputs of expert MLPs. In training and test phase of the recognizer, we utilize the multiple dithered numeral images and the combination of the outputs for corresponding dithered images. Experimental results show that our recognition method works very well.