• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.11 no.7
• /
• pp.3594-3607
• /
• 2017

The traditional feature extraction methods such as principal component analysis (PCA) cannot obtain the local structure of the samples, and locally linear embedding (LLE) cannot obtain the global structure of the samples. However, a common drawback of existing PCA and LLE algorithm is that they cannot deal well with the sparse problem of the samples. Therefore, by integrating the globality of PCA and the locality of LLE with a sparse constraint, we developed an improved and unsupervised difference algorithm called Sparse Difference Embedding (SDE), for dimensionality reduction of high-dimensional data in small sample size problems. Significantly differing from the existing PCA and LLE algorithms, SDE seeks to find a set of perfect projections that can not only impact the locality of intraclass and maximize the globality of interclass, but can also simultaneously use the Lasso regression to obtain a sparse transformation matrix. This characteristic makes SDE more intuitive and more powerful than PCA and LLE. At last, the proposed algorithm was estimated through experiments using the Yale and AR face image databases and the USPS handwriting digital databases. The experimental results show that SDE outperforms PCA LLE and UDP attributed to its sparse discriminating characteristics, which also indicates that the SDE is an effective method for face recognition.

• Journal of Information Processing Systems
• /
• v.14 no.5
• /
• pp.1114-1135
• /
• 2018

A multiple classification system based on a new boosting technique has been approached utilizing different biometric traits, that is, color face, iris and eye along with fingerprints of right and left hands, handwriting, palm-print, gait (silhouettes) and wrist-vein for person authentication. The images of different biometric traits were taken from different standard databases such as FEI, UTIRIS, CASIA, IAM and CIE. This system is comprised of three different super-classifiers to individually perform person identification. The individual classifiers corresponding to each super-classifier in their turn identify different biometric features and their conclusions are integrated together in their respective super-classifiers. The decisions from individual super-classifiers are integrated together through a mega-super-classifier to perform the final conclusion using programming based boosting. The mega-super-classifier system using different super-classifiers in a compact form is more reliable than single classifier or even single super-classifier system. The system has been evaluated with accuracy, precision, recall and F-score metrics through holdout method and confusion matrix for each of the single classifiers, super-classifiers and finally the mega-super-classifier. The different performance evaluations are appreciable. Also the learning and the recognition time is fairly reasonable. Thereby making the system is efficient and effective.

• Annual Conference of KIPS
• /
• 2020.11a
• /
• pp.1000-1003
• /
• 2020

본 논문은 기계학습 기반 온라인 한글 필기 인식 시스템의 첫 구현 결과를 담고 있다. 한글의 글자는 최소한 하나의 모음을 포함하고 있으며, 이 모음은 대개 직선으로 필기한다는 사전 지식을 활용하여 인식에 적용하고자 한다. 이를 위해 사용자가 온라인으로 필기하면 획 데이터를 획득하여 중성에 해당하는 모음을 찾는 알고리즘을 개발하였다. 제안한 알고리즘에서는, 우선 필기한 글자를 포함하는 사각형 R과 각 획을 둘러싸는 사각형 SR을 생성한 후, 직선을 판별하고, 이 직선들이 모음을 구성하는 후보군을 찾는 과정으로 구성되어 있다. 아직 초기 연구이므로, 다양한 경우에 대한 분석이나 실험 결과는 없지만, 이를 활용하여 온라인 필기 인식 모델에 적용하여 인식 성능을 높이기 위한 추후 연구의 기반으로 활용하고자 한다.

• Journal of the Korean Institute of Telematics and Electronics T
• /
• v.35T no.3
• /
• pp.53-60
• /
• 1998

In this paper, we proposed a set of ANSI-Korean character patterns for handwriting recognition that can be used as an input method of mobile computers like PDA (Personal Digital Assistant). In the case of bilinguals, two kinds of alphabets are written alternatively So the method of input character mode change must be provided, and this cause discomfort of writing. Our proposed written character patterns have some constraint but permit ANSI-Korean mixed writing without mode change keeping original form of alphabets and can be recognized with simple algorithm relatively. For ANSI character we analysis Graffiti and propose new writing pattern, which is more similar to original form. There are many researches about input method of unpacking Korean character and writing patterns. But they are not widely used because it's excessively contrary to original form of Korean characters. To show our proposed writing patterns usefulness, we studied the satisfaction and easiness of writing and the recognition rates. Writers are divided into two groups; PDA users, familiar to Graffiti, and others. The results satisfy usefulness in the both groups.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.10
• /
• pp.1271-1279
• /
• 2020

Text region detection is a technology that detects text area in handwriting or printed documents. The detected text areas are digitized through a recognition step, which is used in various fields depending on the purpose of use. However, the detection result of the small text unit is not suitable for the industrial field. In addition, the border of tables in the document that it causes miss-detected results, which has an adverse effect on the recognition step. To solve the issues, we propose a method for detecting text region using the border information of the table. In order to utilize the border information of the table, the proposed method adjusts the flow of two decoders. Experimentally, we show improved performance using the table border pseudo label based on weak supervised learning.

• The KIPS Transactions:PartB
• /
• v.9B no.4
• /
• pp.481-490
• /
• 2002

Off-line handwritten character recognition is in difficulty of incomplete preprocessing because it has not dynamic information has various handwriting, extreme overlap of the consonant and vowel and many error image of stroke. Consequently off-line handwritten character recognition needs to study about preprocessing of various methods such as binarization and thinning. This paper considers running time of watershed algorithm and the quality of resulting image as preprocessing for off-line handwritten Korean character recognition. So it proposes application of effective watershed algorithm for segmentation of character region and background region in gray level character image and segmentation function for binarization by extracted watershed image. Besides it proposes thinning methods that effectively extracts skeleton through conditional test mask considering routing time and quality of skeleton, estimates efficiency of existing methods and this paper's methods as running time and quality. Average execution time on the previous method was 2.16 second and on this paper method was 1.72 second. We prove that this paper's method removed noise effectively with overlap stroke as compared with the previous method.

• Journal of Intelligence and Information Systems
• /
• v.11 no.1
• /
• pp.1-15
• /
• 2005

As organizations search fur more secure authentication methods (Dr user access, e-commerce, and other security applications, biometrics is gaining increasing attention. Biometrics offers greater security and convenience than traditional methods of personal recognition. In some applications, biometrics can replace or supplement the existing technology. In others, it is the only viable approach. Several biometric methods of identification, including fingerprint hand geometry, facial, ear, iris, eye, signature and handwriting have been explored and compared in this paper. They all are well suited for the specific application to their domain. This paper briefly identifies and categorizes them in particular domain well suited for their application. Some methods are less intrusive than others.

• Journal of Internet Computing and Services
• /
• v.6 no.1
• /
• pp.105-114
• /
• 2005

This paper describes a new measuring technique by analysing multiple time-series patterns. This paper's goal is that extracts a really measured value having a sample pattern which is the best matched with an inputted time-series, and calculates a difference ratio with the value. Therefore, the proposed technique is not a recognition but a measurement. and not a hardware but a software. The proposed technique is consisted of three stages, initialization, learning and measurement. In the initialization stage, it decides weights of all parameters using importance given by an operator. In the learning stage, it classifies sample patterns using LBG and DTW algorithm, and then creates code sequences for all the patterns. In the measurement stage, it creates a code sequence for an inputted time-series pattern, finds samples having the same code sequence by hashing, and then selects the best matched sample. Finally it outputs the really measured value with the sample and the difference ratio. For the purpose of performance evaluation, we tested on multiple time-series patterns obtained from etching machine which is a semiconductor manufacturing.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.43 no.5 s.311
• /
• pp.37-44
• /
• 2006

In this paper, a new version of the ICM method(MICM, modified ICM) in which the contextual information is modelled by Markov random fields (MRF) is introduced. To extract the feature, a new local MRF model with a fitting block neighbourhood is proposed. This model selects contextual information not only from the relative intensity levels but also from the geometrically directional position of neighbouring cliques. Feature extraction depends on each block's contribution to the local variance. They discriminates it into several regions, for example context and background. Boundaries between these regions are also distinctive. The proposed algerian performs segmentation using directional block fitting procedure which confines merging to spatially adjacent elements and generates a partition such that pixels in unified cluster have a homogeneous intensity level. From experiment with ink rubbed copy images(Takbon, 拓本), this method is determined to be quite effective for feature identification. In particular, the new algorithm preserves the details of the images well without over- and under-smoothing problem occurring in general iterated conditional modes (ICM). And also, it may be noted that this method is applicable to the handwriting recognition.

• The Journal of the Korea Contents Association
• /
• v.11 no.10
• /
• pp.41-48
• /
• 2011

In feature selection, the selective evaluation scheme for Ant Colony Optimization(ACO) has recently been proposed, which reduces computational load by excluding unnecessary or less promising candidate solutions from the actual evaluation. Its superiority was supported by experimental results. However the experiment seems to be not statistically sufficient since it used only one dataset. The aim of this paper is to analyze convergence characteristics of the selective evaluation scheme and to make the conclusion more convincing. We chose three datasets related to handwriting, medical, and speech domains from UCI repository whose feature set size ranges from 256 to 617. For each of them, we executed 12 independent runs in order to obtain statistically stable data. Each run was given 72 hours to observe the long-time convergence. Based on analysis of experimental data, we describe a reason for the superiority and where the scheme can be applied.