• Journal of KIISE:Software and Applications
• /
• v.32 no.6
• /
• pp.570-579
• /
• 2005

Facial expression recognition technology that has potentialities for applying various fields is appling on the man-machine interface development, human identification test, and restoration of facial expression by virtual model etc. Using sequential facial images, this study proposes a simpler method for detecting human facial expressions such as happiness, anger, surprise, and sadness. Moreover the proposed method can detect the facial expressions in the conditions of the sequential facial images which is not rigid motion. We identify the determinant face and elements of facial expressions and then estimates the feature regions of the elements by using information about color, size, and position. In the next step, the direction patterns of feature regions of each element are determined by using optical flows estimated gradient methods. Using the direction model proposed by this study, we match each direction patterns. The method identifies a facial expression based on the least minimum score of combination values between direction model and pattern matching for presenting each facial expression. In the experiments, this study verifies the validity of the Proposed methods.

• The Journal of the Korea Contents Association
• /
• v.6 no.6
• /
• pp.33-41
• /
• 2006

Nowadays, CCTV can be come across easily in public institutions, banks, and etc. These CCTV play very important roles for preventing many kinds of crimes and resolving those crime affairs. But in the case of recording image of a specific person far from the CCTV, the original image needs to be enlarged and recovered in order to identify the person more obviously. Interpolation is usually used for the enlargement and recovery of the image in this case. However, it has a certain limitation. As the magnification of enlargement is getting bigger, the quality of the original image can be worse. This paper uses FDP(Facial Definition Parameter) proposed by the MPEG-4 SNHC FBA group and introduces a new algorithm that uses face outline information of the original image based on the FDP, which makes it possible to recover better than the known methods until now.

• The Journal of the Institute of Internet, Broadcasting and Communication
• /
• v.17 no.5
• /
• pp.1-8
• /
• 2017

Recently, with the development of ICT technology such as cloud and mobile, image utilization through social networks is increasing rapidly. These images contain personal information, and personal information leakage accidents may occur. As a result, studies are underway to recognize and mask personal information in images. However, optical character recognition, which recognizes personal information in images, varies greatly depending on brightness, contrast, and distortion, and Korean recognition is insufficient. Therefore, in this paper, we design and implement a personal information identification and masking system based on image recognition through deep learning application using CNN algorithm based on optical character recognition method. Also, the proposed system and optical character recognition compares and evaluates the recognition rate of personal information on the same image and measures the face recognition rate of the proposed system. Test results show that the recognition rate of personal information in the proposed system is 32.7% higher than that of optical character recognition and the face recognition rate is 86.6%.

• Journal of the Institute of Electronics Engineers of Korea CI
• /
• v.44 no.3
• /
• pp.98-107
• /
• 2007

For high-dimensional pattern recognition, such as face classification, the small number of training samples leads to the Small Sample Size problem when the number of pattern samples is smaller than the number of dimensionality. Recently, various LDA-extensions have been developed, including LDA, PCA+LDA, and Direct-LDA, to address the problem. This paper proposes a method of improving the classification efficiency by increasing the number of (sub)-classes through pre-clustering a training set prior to the execution of Direct-LDA. In LDA (or Direct-LDA), since the number of classes of the training set puts a limit to the dimensionality to be reduced, it is increased to the number of sub-classes that is obtained through clustering so that the classification performance of LDA-extensions can be improved. In other words, the eigen space of the training set consists of the range space and the null space, and the dimensionality of the range space increases as the number of classes increases. Therefore, when constructing the transformation matrix, through minimizing the null space, the loss of discriminatve information resulted from this space can be minimized. Experimental results for the artificial data of X-OR samples as well as the bench mark face databases of AT&T and Yale demonstrate that the classification efficiency of the proposed method could be improved.

• Proceedings of the Korea Contents Association Conference
• /
• 2004.11a
• /
• pp.333-338
• /
• 2004

Today CCTV can be come across easily in public institutions, banks and etc. These CCTV plays very important roles for preventing many kinds of crimes and resolving those crime affairs. But in the case of recording a image of a specific person far from the CCTV, the original image needs to be enlarged and recovered in order to identify the person more obviously. The interpolation is usually used for the enlargement and recovery of the image. This interpolation has a certain limitation. As the magnification of enlargement is getting bigger, the quality of the original image can be worse than before. This paper uses FDP(Face Definition Parameter) of MPEG-4 SNHC FBA group and introduces a new algorithm that the face outline of a face image using Vector Descriptor based on the FDP makes possible better image recovery than the known methods until now.

• Review of KIISC
• /
• v.25 no.4
• /
• pp.43-50
• /
• 2015

바이오인식기술은 사람의 지문 얼굴 홍채 정맥 등 신체적 특징(Physiological characteristics) 또는 음성 서명 자판 걸음걸이 등 행동적 특징(Behavioral characteristics)을 자동화된 IT 기술로 추출 저장하여 다양한 IT 기기로 개인의 신원을 확인하는 사용자 인증기술이다. 2001년 미국의 911 테러사건으로 인하여 전 세계 국제공항 항만 국경에서 지문 얼굴 홍채 등 바이오정보를 이용한 출입국심사가 보편화됨과 동시에 ISO/IEC JTC1 SC37(바이오인식) 국제표준화기구를 중심으로 표준화가 급속도로 진행되어 왔다. 최근 들어 스마트폰 테블릿 PC 등 모바일기기에 지문 얼굴 등 바이오정보를 탑재하여 다양한 모바일 응용서비스를 가능하게 해주는 모바일 바이오인식 응용기술이 전 세계적으로 개발 보급되고, 삼성전자 페이팔 중심으로 바이오인식기술을 이용한 모바일 지급결제솔루션에 대한 사실표준화협의체인 FIDO, ITU-T SG17 Q9(텔레바이오인식) 국제표준화기구를 중심으로 표준화가 진행되고 있다. 특히 이러한 모바일 바이오인식기술은 스마트폰을 통한 비대면 인증기술 수단으로서 핀테크의 중요한 요소기술로 작용될 전망이다. 한편, 위조지문 등 전통적인 바이오인식 기술의 위변조 위협으로 인한 우려도 증폭됨에 따라 스마트워치 등 웨어러블 디바이스에서 살아있는 사람의 심박수(심전도), 뇌파 등의 생체신호를 측정하여 스마트폰을 통하여 개인을 식별하는 차세대 바이오인식기술로 진화중에 있다. 본고에서는 바이오인식기술의 변천사와 함께 국내외 모바일 바이오인식기술 동향과 표준화 추진현황을 살펴보고, 지난 2015년 5월 29일 발족한 KISA "모바일 생체신호 인증기술 표준연구회"를 통하여 뇌파 심전도 등생체신호를 이용한 차세대 바이오인식 기술 및 표준화 계획을 수립하여 향후 바이오인식기반의 비대면 인증기술에 대한 추진전략을 모색하고자 한다.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.5
• /
• pp.1-8
• /
• 2021

Coronavirus disease 2019 (COVID-19) has affected the world seriously. Every person is required for wearing a mask properly in a public area to prevent spreading the virus. However, many people are not wearing a mask properly. In this paper, we propose an efficient mask detection system. In our proposed system, we first detect the faces of input images using YOLOv5 and classify them as the one of three scene complexity classes (Simple, Moderate, and Complex) based on the number of detected faces. After that, the image is fed into the Faster-RCNN with the one of three ResNet (ResNet-18, 50, and 101) as backbone network depending on the scene complexity for detecting the face area and identifying whether the person is wearing the mask properly or not. We evaluated our proposed system using public mask detection datasets. The results show that our proposed system outperforms other models.

• Journal of Convergence for Information Technology
• /
• v.9 no.7
• /
• pp.100-105
• /
• 2019

As IoT Technology develops and Era of Hyperconnectivity comes, various kinds of customized services became available. As a next-generation display, a smart mirror accesses multimedia devices and provides various services, so it can serve as a social learning tool for the children and the old ones, as well as adults who need information. Smart Mirror must be able to identify users for individualized services. However, since the Smart Mirror is an easily accessible device, there is a possibility that information such as an individual's pattern and habit stored in the smart mirror may be exposed to the outside. Also, the other possibility of leakage of personal location information is through personal schedule or appointment stored in the smart mirror, and another possibility that privacy can be violated is through checking the health state via personal photographs. In this research, we propose a system that identify users by the information the users registered about their physique just like their face, one that provides individually customized service to users after identifying them, and one which provides minimal information and service for unauthenticated users.

• Science of Emotion and Sensibility
• /
• v.19 no.1
• /
• pp.95-110
• /
• 2016

Uncanny valley refers to the condition where the affinity of a human-like object decreases dramatically if the object becomes extremely similar to human, and has been hypothesized to derive from the cognitive load of categorical conflict against an uncanny object. According to the hypothesis, the present study ran an oddball task consisting of trials each displaying one among a non-human, human and uncanny face, and measured event-related potentials (ERPs) for each trial condition. In Experiment 1, a non-human face was presented in 80% of the trials (standard) whereas a human face for another 10% trials (target) and an uncanny face for the remaining 10% trials (uncanny). Participants' responses were relatively inaccurate and delayed in both the target and uncanny oddball trials, but neither P3 nor N170 component differed across the three trial conditions. Experiment 2 used 3-D rendered realistic faces to increase the degree of categorical conflict, and found the behavioral results were similar to Experiment 1. However, the peak amplitude of N170 of the target and uncanny trials were higher than the standard trials while P3 mean amplitudes for both the target and uncanny trials were comparable but higher than the amplitude for the standard trials. P3 latencies were delayed in the order of the standard, target, and uncanny trials. The changes in N170 and P3 patterns across the experiments appear to arise from the categorical conflict that the uncanny face must be categorized as a non-target according to the oddball-task requirement despite its perceived category of a human face. The observed increase of cognitive load following the added reality to the uncanny face also indicates that the cognitive load, supposedly responsible for the uncanny experience, would depend on the increase of categorical conflict information subsequent to added stimulus complexity.

• Journal of the Korea Convergence Society
• /
• v.13 no.4
• /
• pp.45-51
• /
• 2022

Fear of contact exists due to the prevention of the spread of infectious diseases such as COVID-19. When using the common entrance door of an apartment, access is possible only if the resident enters a password or obtains the resident's permission. There is the inconvenience of having to manually enter the number and password for the common entrance door to enter. Also, contactless entry is required due to COVID-19. Due to the development of ICT, users can be easily identified through the development of face recognition and voice recognition technology. The proposed method detects a visitor's face through a CCTV or camera attached to the common entrance door, recognizes the face, and identifies it as a registered resident. Then, based on the registered information of the resident, it is possible to operate without contact by interworking with the elevator on the server. In particular, if face recognition fails with a hat or mask, the visitor is identified by voice or additional authentication of the visitor is performed based on the voice message. It is possible to block the spread of contagiousness without leaving any contactless function and fingerprint information when entering and exiting the front door of an apartment house, and without the inconvenience of access.