Earticle

최근 생성형 인공지능의 발전은 학술 콘텐츠 생성 방식에 큰 변화를 가져오고 있으며, 특히 논문 작성 자동화에 대한 수요가 증가하고 있다. 본 연구는 논문작성 템플릿을 기반으로 생성형 인공지능(ChatGPT, Gemini, Copilot) 을 활용하여 자동 논문 생성 시스템을 설계하고 성능을 비교하였다. 연구는 템플릿 기반 구조를 정의하고, 각 AI 모델에 동일한 프롬프트를 입력하여 결과물을 도출한 후 속도, 정확도, 인식율(F1 Score), 사용자 평점 등 다양한 지표로 성능을 분석하였다. 실험 결과, 각 모델은 고유한 특성과 성능 차이를 보였으며, 템플릿 기반 접근 방식은 일관된 구조와 품질 향상에 기여하였다. 본 연구는 학술 콘텐츠 자동화의 효율성을 높이고, 연구자들이 보다 창의적인 작업에 집중할 수 있는 기반을 제공할 수 있다. 향후 다양한 분야에 적용 가능한 템플릿 확장 및 사용자 맞춤형 생성 기능 개발이 기대된다.

Recently, the advancement of generative AI has significantly transformed academic content creation, especially increasing the demand for automated paper writing. This study designs a template-based automatic paper generation system using generative AI models (ChatGPT, Gemini, Copilot) and compares their performance. The research defines a structured template, inputs identical prompts into each AI model, and analyzes the results based on speed, accuracy, F1 Score, and user ratings. Experimental results show distinct characteristics and performance differences among the models, and the template-based approach contributes to consistent structure and improved quality. This study enhances the efficiency of academic content automation and provides a foundation for researchers to focus on more creative tasks. Future work includes expanding templates for various fields and developing personalized generation features.

Interest is increasing in electrocardiogram (ECG) signal analysis for embedded devices, creating the need to develop an algorithm suitable for a low-power, low-memory embedded device. Linear approximation of the ECG signal facilitates the detection of fiducial points by expressing the signal as a small number of vertices. However, dynamic programming, a global optimization method used for linear approximation, has the disadvantage of high complexity using memoization. In this paper, the calculation area and memory usage are improved using a linear approximated template. The proposed algorithm reduces the calculation area required for dynamic programming through local optimization around the vertices of the template. In addition, it minimizes the storage space required by expressing the time information using the error from the vertices of the template, which is more compact than the time difference between vertices. When the length of the signal is L, the number of vertices is N, and the margin tolerance is M, the spatial complexity improves from O(NL) to O(NM). In our experiment, the linear approximation processing time was 12.45 times faster, from 18.18 ms to 1.46 ms on average, for each beat. The quality distribution of the percentage root mean square difference confirms that the proposed algorithm is a stable approximation.

Extensive research has been conducted on template-based method, particularly those using cylindrical channels (e.g., anodized aluminum oxide (AAO) template) to create nanorods since their first introduction in the 1990s. While this approach offers many advantages, it also presents significant practical challenges. One major issue is the tendency of nanorods to aggregate, which negates their individual properties. To preserve these unique properties, nanorods must remain stably dispersed in solution. In a previous study (J. Korean Phys. Soc. 85 (2024) 360-370), nanorods with a diameter of 45 nm, produced using a template with an interpore distance (IPD) of 110 nm, were successfully dispersed under conditions predicted by DLVO (Derjaguin-Landau-Verwey-Overbeek) theory. In this paper, we extend the investigation by exploring a broader range of IPD and diameter values. Our findings show that the nanorod diameter should be approximately 30% of the IPD for stable dispersion.

클라우드 환경의 보안 위협 증가로 최소 권한 정책의 중요성이 부각되고 있다. 특히, AWS와 같은 퍼블릭 클라우드 플랫폼에서는 복잡한 권한 관리로 인한 과도한 권한 부여가 보안 사고로 이어질 수 있다. 기존 최소 권한 정책 데이터베이스는 단순한 호출 기록에 기반하여 종속 권한 누락과 동적 권한 생성의 비효율성을 겪었다. 이를 해결하기 위해 AWS CloudTrail 로그를 기반으로 동적으로 최소 권한 정책을 생성하는 체계를 제안한다. 구체적으로, Action 그룹화, Placeholder 기반 동적 정책 생성, 종속 권한 반영의 세 가지 접근법을 통해 데이터베이스를 최적화하였다. 연구의 결과, AWS의 주요 서비스인 S3와 EC2를 대상으로 총 644개의 API 정책 파일을 설계하고 데이터베이스화하여 실제 클라우드 환경에서도 적용할 수 있는 최소 권한 정책 체계를 구축하였다.

With increasing security threats in cloud environments, the least privilege has become important. In platforms like AWS, complex permission management often leads to over-provisioning and vulnerabilities. Traditional policy databases miss dependent and dynamic permissions due to simple records. To address these challenges, we propose a dynamic framework using CloudTrail logs with action grouping, placeholder-based generation, and dependent permission integration. Thus, we compiled a database of 644 policies for S3 and EC2, creating a scalable least privilege system.

In this paper, we present a noise generation algorithm for producing artistic styles on video. Our algorithm is constructed based on a template to preserve the coherence of the artistic styles between the frames of video. We generate noise in the coordinate defined by the templates and apply texture coverage scheme to control the coherence of the noise distribution. For the coherence of the regions outside the templates, we design local recursive filters.

전화 기능과 컴퓨터 기능을 갖는 스마트폰의 보급으로 장소에 구애받지 않는 모바일 상거래에 대한 수요가 급증하고 있다. 상거래 서비스는 인증 기법을 적용하여 거래 내용과 거래 당사자에 대한 법적 구속력을 갖는 증거를 남겨야 한다. 스마트폰은 개인용 컴퓨터와 달리 개인정보 노출의 위험이 있고 기기 분실 및 도용에 따른 대리 인증이 상대적으로 쉽다. 기존의 패스워드 및 토큰 기반의 인증 기법은 사용자와 아이디를 물리적으로 연결시키지 못하므로 대리 인증 문제를 해결할 수 없다. 따라서 스마트폰을 모바일 상거래 플랫폼으로 활용하기 위해서는 법적 구속력이 있고 대리 인증이 어려운 새로운 사용자 인증 기법에 대한 연구가 필요하다. 본 논문에서는 사용자 고유 음성 정보와 스마트폰 USIM 정보를 접목한 모바일 ID를 생성하고 이에 기반을 둔 사용자 인증 기법을 설계 및 구현하였다.

A smart phone has functions of both telephone and computer. The wide spread use of smart phones has sharply increased the demand for mobile commerce. The smart phone based mobile services are available anytime, anywhere. In commercial transactions, a digital signature scheme is used to make legally binding signature to prove both integrity of commercial document and verification of the signer. Smart phones are more risky compared with personal computers on the problems of how to protect privacy information. It’s also easy to let proxy user to authenticate instead of the smart phone owner. In existing password or token based schemes, the ID is not physically bound to the owner. Thus, those schemes can not solve the problem of proxy authentication. To utilize the smart phone as the platform of mobile commerce, a study on the new type of authentication scheme is needed where the scheme should provide protocol to get legally binding signature and not to authenticate proxy user. In this paper, we create the mobile ID by using both the USIM and voice template of the smart phone owner. We also design and implement the user authentication scheme based on the mobile ID.

본 연구는 PCA와 템플릿 정합을 사용한 얼굴 표정 인식 알고리즘을 제안한다. 먼저 얼굴 영상은 Haar-like feature의 특징 마스크를 사용하여 획득한다. 획득한 얼굴 영상은 눈과 눈썹을 포함하고 있는 얼굴 상위 부분과 입과 턱을 포함하고 있는 얼굴 하위 부분으로 분리하여 얼굴 요소 추출에 용이하게 나눈다. 얼굴 요소 추출은 눈 영상과 입 영상을 추출하는 과정으로 먼저 학습영상으로 PCA를 거쳐 생성된 고유얼굴을 구한다. 고유 얼굴에서 고유 입과 고유 눈을 획득하고, 이를 얼굴 분리 영상과 템플릿 매칭시켜 얼굴요소를 추출한다. 얼굴 요소는 눈과 입이 있으며 두 요소의 기하학적 특징으로 표정을 인식한다. 컴퓨터 모의실험 결과에 따르면 제안한 방법이 기존의 방법보다 추출률이 우수하게 나왔으며, 특히 입 요소의 추출률은 99%에 달하였다. 또 이 얼굴 요소 추출 방법을 표정인식에 적용하였을 때 놀람, 화남, 행복의 3가지 표정의 인식률이 80%를 상회하였다.

This paper proposed a recognition algorithm of human facial expressions using the PCA and the template matching. Firstly, face image is acquired using the Haar-like feature mask from an input image. The face image is divided into two images. One is the upper image including eye and eyebrow. The other is the lower image including mouth and jaw. The extraction of facial components, such as eye and mouth, begins getting eye image and mouth image. Then an eigenface is produced by the PCA training process with learning images. An eigeneye and an eigenmouth are produced from the eigenface. The eye image is obtained by the template matching the upper image with the eigeneye, and the mouth image is obtained by the template matching the lower image with the eigenmouth. The face recognition uses geometrical properties of the eye and mouth. The simulation results show that the proposed method has superior extraction ratio rather than previous results; the extraction ratio of mouth image is particularly reached to 99%. The face recognition system using the proposed method shows that recognition ratio is greater than 80% about three facial expressions, which are fright, being angered, happiness.

본 논문에서 호흡 기반의 운전자 졸음 감지 센서 시스템에 대해 언급하였다. 센서 시스템은 운전자의 복부 부분 안 전벨트에 장착된 PZT 압력센서와 개인용 컴퓨터로 구성됐다. PZT 압력센서는 호흡 시 운전자 복부의 움직임에 의해 압력센서에 가해지는 압력의 변화를 측정하기 위해 사용되었고 운전자의 졸음을 감지하기 위한 신호처리는 Labview를 사용하여 개발됐다. 30세 남자 운전자를 상대로 운전자 졸음 감지 관련 실험들이 수행되어 졌다. 운전자가 각성상태일 때 호흡의 크기는 졸음상태일 때보다 컸으며 반대로 호흡 주파수는 낮았다. 이런 실험을 바탕으로 제작된 졸음 감지 센서 시스템은 운전자의 졸음을 성공적으로 실시간 감지할 수 있었다.

In this paper, a driver's drowsy detection sensor system based on the respiration is investigated. The sensor system consists of a piezoelectric pressure sensor attached at the abdominal region of the seat belt and a personal computer. The piezoelectric pressure sensor was utilized for the measurement of pressure variations induced by the movement of the driver abdomen during breathing. The signal processing software for detecting driver's drowsiness was produced using the Labview. The experiments were performed with 30 years male driver. The amplitude of the respiration at awake state was larger than one at the drowsy state. On the contrary, the respiration rate at awake state was lower than one at the drowsy state. The drowsy detection sensor system developed based on the experimental could successfully detect the driver's drowsy on real-time.

Nowadays main infrastructure-as-a-service (IaaS) systems have been widely exploiting the template-based VM creation and template caching techniques to reduce the startup latency of user VM and service response time. Because the new VMs created from the templates do not always have the same software system as needed, further reconfigurations, that is, installing the missing software components and removing the undesired ones, must be made before the VMs can be used by users. The reconfiguration time is also an important source of the VM startup latency. However, current template caching solutions based on the traditional caching strategy select some templates which are frequently used recently to cache without considering the reconfiguration time of various user VMs created in IaaS data center. In this paper, we exploit a modified k-means clustering method to optimize the selecting of templates to be cached and theoretically prove the selected ones can minimize the average reconfiguration time and then the average startup latency of all user VMs under the same limits on the number of templates which cache space can accommodate. The correlative simulation experiments also prove the effectiveness of our approach.

For improving automation calibration precision of multiple view texture collection in texture reconstruction, in this paper a new kind of calibration template using statistical distribution of surface texture is proposed. Firstly, there is a normalization processing be made according to statistical distribution of texture feature, such as species of color, quantity of geometric pattern and bump texture, etc. for the division of viewpoint. Then, the camera views of calibration template are assigned by using the normalization data with the calibration algorithm. Finally, the algorithm analysis is made by viewpoint division of four cases and the accuracy of our method can improve more than 10 percent. The experimental results show that, the proposed new calibration template method can better improve precision of 2D-3D registration and texture accuracy of reconstruction model than the uniform calibration template, while no obviously increasing the amount of calculation.

Object tracking is an important function of video surveillance system. For the same object in a multi-camera environment, how to assign the same label to this object is so-called consistent labeling problem. Many consistent labeling approaches proposed by previous studies mainly based on the environment with multiple fixed cameras. In this study, a consistent labeling approach is proposed for a PTZ (Pan-Tilt-Zoom) camera. The same object is assigned the same label while the object in the FOV (Field-of-View) of a PTZ camera without influencing by the pan/tilt rotation. In order to achieve the above goal, the proposed approach using several methods, such as temporal differencing, template matching, mean-shift tracking, Kalman filter, and so on. A template cache is also designed for preserving the templates of an object with various angles and a least recently used (LRU) replacement mechanism is used to update the cache. The experimental results show that accuracy of the proposed approach for the consistent labeling of a PTZ camera can reach about 83 percentage.

In the face recognition system, the protection of the facial feature template has already become the most important part. In view of that the current protection methods cannot meet the security and the matching rate simultaneously, a hybrid encryption algorithm was proposed in this paper fusing chaotic encryption and RSA encryption to solve this problem. After the preprocessing to the ORL face database, the feature template was extracted by 2D2LDA and PCA in the non-transformation domain. Then we applied the scrambling encryption and RSA encryption to obtain the final security template. At last, a system with the nearest neighbor classifier and Euclidean distance was used for the matching verification. The experimental results showed that this hybrid encryption method was feasible, effective and easy transplantation with no matching rate reduction.

The Project presents object tracking from videos based on template matching using Gaussian kernel with probability distribution function and mean shift algorithm. Initially, the target will be selected from chosen video sequence to track desired object in consecutive frames. The target will be utilized to determine the probability distribution function for similarity measurement between target and current processing frames. Here, Gaussian kernel function and its gradient are used here to find the PDF for corresponding templates. Similarity between two different images will be measured by weighted sum of Gaussian coefficients and PDFS. Mean shift approach used here to shifting the starting coordinates of template to find its similar features in consecutive frames to detect desired objects. The dissimilarity between the target model and target candidates will be expressed by a metric derived from Bhattacharyya coefficient. The project simulated results shows that moving object from video will be tracked accurately at different position and shape with help of templates in a considerable amount of time.

Fischer Tropsch reaction is one of the interesting topic for renewable and clean energy. Polymerization of carbon monoxide or carbon dioxide with hydrogen over metal supported catalyst can produce long chain hydrocarbons. Synthetic liquid hydrocarbons are promising alternative to fossil fuels. This research work has been focused on the synthesis of Fe based catalyst for Fischer Tropsch reaction. Mesoporous silica (MS) support prepared by a precipitation method using two different washing solution, distilled water (DW) and acid in ethanol solution (ET), and different calcination temperature. Then, Fe/MS was prepared by an incipient wetness impregnation method. All of samples were systematically characterized using various physical and chemical techniques. TEM and XRD analysis were used to ensure that the cubic Ia3d mesostructure is stable after calcination. FTIR spectra are useful to ascertain the existence of template in the support. TPR studies were also used to understand the nature of Fe species and their reducibility. The results reveal that washing the support with distilled water and calcination at 550 can efficiently remove the triblock copolymer templates. The existence of template in the support affects the textural properties of all catalyst investigated.

Finite mixture model (FMM) with Gaussian distribution has been widely used in many image processing and pattern recognition tasks. This paper presents a new Student's-t mixture model (SMM) based on Markov random field (MRF) and weighted mean template. In this model, the Student's-t distribution is considered as an alternative to the Gaussian distribution due to the former is heavily tailed than Gaussian distribution, thus providing robustness to outliers. With the help of the weighted mean template, the spatial information between neighboring pixels of an image is considered during the learning step. In addition, the proposed method is able to impose the smoothness constraint on the pixel label by using MRF. Furthermore, an efficient energy function and a novel factor are applied in current model to decrease the computational complexity. Numerical experiments are presented on simulated and real world images, and the results are compared with other FMM-based models.

With the explosive growth of the information on the Internet, the number of web pages is also expanding rapidly. A common web page typically contains many different blocks. Unfortunately, apart from the main content blocks, it usually has such blocks as navigation bars, copyright, and advertisements, which are known as noise. Studies have shown that these noisy blocks can do harm to Web mining, search engine performance, and Web page clustering and classification. It is a significant work to effectively clean these noises up and extract main content blocks from the Web pages. In this paper, we propose a method of automatic template detection and noise elimination based on an new idea defining two sorts of templates called Style Template and Content Template relatively. There is a basic observation that, the Web pages from the same site are always similar in presentation style and contents, however, the main content blocks are always different in contents. Therefore, we design an algorithm to detect the two kind of templates from the samplings from a particular site, and extract the main content out of the noises using the template. We apply our method to template problem, and the result shows that our methods performs very well, and has a remarkable advantage on time consumption over previous methods.

Rapid diagnosis of bacterial infections has been considered as important criteria for a patient management and an appropriate therapy of bacteria-induced disease in the early phase. Among several techniques, CE-SSCP combined with 16S rRNA gene-specific PCR takes a huge attention due to its benefits such as high sensitivity, resolution and reproducibility. However, as in any other PCR-based technology, multiplex detection and quantification is a major hurdle for this method because of primer dimer formation and non-specific amplification by multiple sets of primers. In this study, we developed a novel technique for multiplex detection and quantification of pathogens by CE-SSCP coupled with multiplex asymmetric PCR using common primers via template-tagging. For linearly amplifying the reversed transcripts of 16S rRNAs from nine septicemia-inducing pathogens, the reverse transcripts were tagged with the common sequence and subsequently amplified by asymmetric PCR using the common primers. After the amplification, the amplicons were separated and quantified by CE-SSCP. The results illustrated the potentials of this method in the diagnosis of the infectious disease.

본 논문은 컴퓨터 비전(computer vision) 분야의 영상처리(image processing) 기술 중 물체인식(object recognition)과정에서 사용되는 템플릿 매칭(template matching)을 예술의 영역으로 가져와 건축공간의 연속적 경험을 영상으로 표현하기 위한 시스템에 활용할 수 있음을 제안한다. 템플릿 매칭은 이미지 간의 유사도를 검출해 낼 수 있다. 이러한 특징을 이용하여 특정한 건축공간을 촬영한 수많은 이미지들 중에서 임의로 템플릿을 선택하고, 선택된 템플릿과 유사한 이미지들을 검출하면 연속적인 장면으로 이루어진 하나의 시퀀스를 구성할 수 있다. 이러한 작업을 반복 수행함으로써 공간의 경험을 여러 시퀀스로 구성된 하나의 영상으로 표현할 수 있다. 또한 많은 입력 이미지 중 어떠한 이미지를 템플릿으로 선택하느냐에 따라 그것과 유사한 이미지들이 다르게 검출된다. 따라서 같은 건축공간이라 할지라도 매번 다른 시퀀스로 구성된 영상을 만들 수 있다. 본 논문에서는 공간의 구조, 색상과 같은 건축 이미지들의 특징을 반영하고 장면들을 연속적인 형태로 구성하기 위하여 템플릿 매칭의 다양한 실험을 진행하였다. 그 결과 인간의 눈처럼 완벽하지는 않지만, 공간 안에서 시점의 변화에 따른 연속적인 장면들을 검출해 내어 영상으로 구성할 수 있었다.

This paper suggests that template matching techniques for use in the object recognition process for image processing in the field of computer vision. Applications can be found in the areas of art and systems to visualize continuous experiences in architectural space. Using the template matching technique, the similarity between a template and input images is computed. This can assist with the selection of a random template from numerous images of a particular architectural space and in the finding of images that are similar to the selected template in order to create a sequence of continuous images. By repeating this process, spatial experiences can be represented in a video composed of multiple sequences. In addition, depending on the image selected as a template from numerous images, different images that are similar to the templates can be found. Therefore, every time a template is selected, different videos composed of different sequences can be created. For this study, I conducted a number of template matching experiments in order to find images that reflect the architectural characteristics of a space, such as the spatial structure or colors, after which I compose these images in a continuous format. As a result, I was able to find continuous images that change according to changes in the viewpoint and to create a video out of them, although the video is not as perfect as observation by the human eye.