Earticle

Copy?Move Forgery Detection (CMFD) helps to detect copied and pasted areas in one image. It plays a crucial role in legal evidence, forensic investigation, defence, and many more places. In the proposed CMFD method, a two-step identification of forgery is presented. In step one, the suspected image will be classified into either one of two classes that are forged or authentic. Step two is carried out? only if the suspected is classified as forged, then forged location will be identified using the block-matching procedure. Initially, the suspected image is decomposed into different orientations using Steerable Pyramid Transform (SPT); Grey Level Co-occurrence Matrix (GLCM) features are extracted from each orientation. These features are used to train Optimized Support Vector Machine (OSVM) as well as to classify. If the suspected image is categorized into forged, then the suspected grey image is converted into overlapping blocks, and from each block, GLCM features are extracted. The proper similarity threshold value and distance threshold value can locate the forged region using GLCM block features. The performance of the proposed method is tested using standard datasets CoMoFoD and CASIA Datasets. The proposed CMFD approach results are consistent, even the forged image suffered from attacks like JPEG compression, scaling, and rotation. The OSVM classifier is showing superiority over the Optimized Naive Bayes Classifier (ONBC), Extreme Learning Machine (ELM) and Support Vector Machine (SVM).

의료 사물인터넷(IoMT)환경의 확산과 함께 네트워크 패킷 조작 및 생체 데이터 위·변조를 통한 사이버 위협이 증가하고 있다. 기존 IoMT 침입탐지 연구는 주로 네트워크 feature에 의존하였으며, 설명 가능한 AI 기법인 SHAP을 적용한 연구도 결과의 통계적 타당성 검증 없이 특징(feature) 중요도를 해석하는 방법론적 한계가 존재하였다. 본 연구는 WUSTL-EHMS-2020 데이터셋을 기반으로 XGBoost, Random Forest, KNN 모델을 적용한 멀티클래스 침입탐지 체계를 구축하고, Ablation Study, 통계적 분포 검증, SHAP·LIME XAI 분석으로 구성된 3중 검증 체계를 제안하였다. 실험 결과, XGBoost가 F1-score 0.9548로 최고 성능을 달성하였다. Ablation Study를 통해 생체 feature 통합 시 Spoofing 탐지 F1이 0.7962에서 0.8848로 향상됨을 확인하였으며, 이는 선행연구에서 보고된 AUC 25% 향상의 실질적 기여자가 Spoofing 클래스에서의 생체 feature 시너지임을 멀티클래스 분석으로 정량화한 것이다. 또한 SHAP 분석에서 Spoofing 탐지에 체온이 상위 기여 feature로 나타났으나, t-검정 및 LIME 분석과 의 불일치는 이것이 실제 판별력이 아닌 교란변수 효과임을 시사하였다. 본 연구에서 제안하는 3중 검증체계는 XAI 결과의 신뢰성을 보장하는 방법론적 안전망으로서 IoMT 환경에서 설명 가능하고 검증 가능한 AI 기반 침입탐지 모니터링 체계 구축을 위한 실용적 기반이 될 수 있다.

As the Internet of Medical Things environment continues to expand, cyber threats involving network packet manipulation and biomedical data falsification are increasingly emerging. Existing IoMT intrusion detection studies have primarily relied on network features, and even studies applying explainable AI techniques such as SHAP have faced methodological limitations in interpreting feature importance without statistical validation of the results. This study constructs a multiclass intrusion detection system applying XGBoost, Random Forest, and KNN models based on the WUSTL-EHMS-2020 dataset, and proposes a three-tier validation framework comprising Ablation Study, statistical distribution verification, and dual SHAP·LIME XAI analysis. Experimental results demonstrate that XGBoost achieves the highest performance with an F1- score of 0.9548. Ablation Study confirms that integrating biomedical features improves Spoofing detection F1 from 0.7962 to 0.8848, quantifying that the actual contributor to the 25% AUC improvement reported in prior research is biomedical feature synergy in the Spoofing class through multiclass analysis. Furthermore, although SHAP analysis identifies body temperature as a top contributing feature for Spoofing detection, inconsistencies with t-test results and LIME analysis suggest this reflects a confounding variable effect rather than actual discriminative power. The proposed three-tier validation framework serves as a methodological safeguard ensuring XAI reliability, and provides a practical foundation for building explainable and verifiable AIbased intrusion detection monitoring systems in IoMT environments.

The condition of an inaccessible gear in an operating machine can be monitored using the vibration signal of the machine measured at some convenient location and further processed to unravel the significance of these signals. Demodulation is an important issue in gearbox fault detection. Non-stationary modulating signals increase difficulties of demodulation. Though wavelet packet transform has better time–frequency localization, because of the existence of meshing frequencies, their harmonics, and coupling frequencies generated by modulation, fault detection results using wavelet packet transform alone are usually unsatisfactory. This paper proposes a fault detection method that combines Hilbert transform and machine learning method namely support vector machines (SVMs). The statistical feature vectors from Hilbert transform coefficients are classified using J48 algorithm and the predominant features were fed as input for training and testing SVM and their efficiency in classifying the faults in the Bevel Gear Box was studied.

Falls in elderly is a very serious health problem. For these years, the wearable devices based on tri-axial accelerator has been proven to be an effective way to fall detection. Most current methods for fall detection are based on threshold and machine learning. A approach based on statistical features was proposed to distinguish falls and normal activities of daily living(ADL) in this paper. What is worth mentioning is that Kernel Principal component analysis(KPCA) is firstly used to extract the statistical features from the original 3D data of acceleration, we don’t need to design features specially. The support vector machine (SVM) algorithm and K-Nearest Neighbor(KNN) algorithm are combined for prediction. Finally the validation of the prediction is done to improve the accuracy. Algorithm is mainly conducted on the public databases(UCI). And our method obtained the result is proved to be better compared with the other literature based on this public databases.

There are many medical imaging modalities used for the analysis and cure of various diseases. One of the most important of these modalities is Magnetic Resonance Imaging (MRI). MRI is advantageous over other modalities due to its high spatial resolution and the excellent capability of discrimination of soft tissues. In this paper, an automated classification approach of normal and pathological MRI is proposed. The proposed model three simple stages; preprocessing, feature extraction and classification. Two types of features; color moments and texture features have been considered as main features for the description of brain MRI. A probabilistic classifier based on logistic function has been used for the MRI classification. A standard data set consisting of one hundred and fifty images has been used in the experiments, which was divided into 66% training and 34% testing. The proposed approach gave 98% accurate results for training data set and 94% accurate results for the testing data set. For validation of the proposed approach, 10-Fold cross validation was applied, which gave 90.66% accurate results. The classification capability of probabilistic classifier has been compared with the different state of art classifiers, including Support Vector Machine (SVM), Naïve Bayes, Artificial Neural Network (ANN), and Normal densities based linear classifier.

The automatic handwriting identification is a hot topic in pattern recognition that it has been extensively studied in many languages. A Uyghur handwriting identification technique based on the stroke statistical features is proposed in this paper. Firstly, the handwriting image is preprocessed taking modified methods of grid line removal, noise reduction and thinning. Then novel stroke statistical features are extracted based on the structural character and writing styles of Uyghur handwriting. And this approach respectively achieves a top 1 and top 2 identification rates of 98.66% and 99.78% on the Uyghur handwriting data set from 224 different people. Finally, Comparison analysis of different stroke length and distance measurement method has been conducted through three different kinds of experiments, the optimal stroke length and distance measurement method is determined, and its effectiveness and stability are tested. The stroke statistical features can capture the structural character and writing style of Uyghur handwriting efficiently, and it is suitable for any languages theoretically.

In this paper, lesion areas affected by anthracnose are segmented using segmentation techniques, graded based on percentage of affected area and neural network classifier is used to classify normal and anthracnose affected on fruits. We have considered three types of fruit namely mango, grape and pomegranate for our work. The developed processing scheme consists of two phases. In the first phase, segmentation techniques namely thresholding, region growing, K-means clustering and watershed are employed for separating anthracnose affected lesion areas from normal area. Then these affected areas are graded by calculating the percentage of affected area. In the second phase texture features are extracted using Runlength Matrix. These features are then used for classification purpose using ANN classifier. We have conducted experimentation on a dataset of 600 fruits’ image samples. The classification accuracies for normal and affected anthracnose fruit types are 84.65% and 76.6% respectively. The work finds application in developing a machine vision system in horticulture field.

The history of Korean migration might be traced back to early as 16th century when Korea engaged wars with Japan and China rrespectively and,after being severely defeated, large number of Korean craftsmen, concubine,peasants and even the members of royal family, was taken hostages to Japan and China. Those Koreans are said to create the first Korean diasporic communities in these countries. To Koreans, the history of 16th century was remembered as one of gloomy history along with the occupation of Korean peninsula by Japan in the early 20th century. Historical experiences on Korean international migration thus shed negative images on diaspora which is still wide-spreading among Koreans. This paper aims to interpret small but significant number of Korean overseas. The Ministry of Foreign Affairs and Trade (MOFAT) issues a biannual report on Korean overseas and the recent version published in 2011shows that there are 7,268,771 Korean overseas over the world. The absolute number is small compare to major disporic communities, i.e. China and India. However, when it comes to the ratio per national population, Korean overseas runs second to the Jewish diasporic community. Of them, more than 80%live in three major countries: China, the United States and Japan. Quite contrary, Korean overseas in Africa countries numbers only 11,072 which occupy 0.15% of total Korean overseas. This paper pursues the reasons of insignificant presence of Korean overseas in African countries. Two reasons,negative image of Africa and strong Confucian thinking which stresses ‘Chung’(national royalty) and ‘Hyo’(family piety), are given to explain the phenomenon. I also ask for further research on Korean overseas in Africa,which is virtually non-exist, since most of academic researches on Korean overseas are focused on major diasporic communities.

Objective: The aim of this study was to develop an automated computer-aided diagnostic system for diagnosis of thyroid cancer pattern in fine needle aspiration cytology (FNAC) microscopic images with high degree of sensitivity and specificity using statistical texture features and a Support Vector Machine classifier (SVM). Materials and Methods: A training set of 40 benign and 40 malignant FNAC images and a testing set of 10 benign and 20 malignant FNAC images were used to perform the diagnosis of thyroid cancer. Initially, segmentation of region of interest (ROI) was performed by region-based morphology segmentation. The developed diagnostic system utilized statistical texture features derived from the segmented images using a Gabor filter bank at various wavelengths and angles. Finally, the SVM was used as a machine learning algorithm to identify benign and malignant states of thyroid nodules. Results: The SVMachieved a diagnostic accuracy of 96.7% with sensitivity and specificity of 95% and 100%, respectively, at a wavelength of 4 and an angle of 45. Conclusion: The results show that the diagnosis of thyroid cancer in FNAC images can be effectively performed using statistical texture information derived with Gabor filters in association with an SVM.

We apply independent component analysis (ICA) for extracting an optimal basis to the problem of finding efficient features for representing speech signals of a given speaker The speech segments are assumed to be generated by a linear combination of the basis functions, thus the distribution of speech segments of a speaker is modeled by adapting the basis functions so that each source component is statistically independent. The learned basis functions are oriented and localized in both space and frequency, bearing a resemblance to Gabor wavelets. These features are speaker dependent characteristics and to assess their efficiency we performed speaker identification experiments and compared our results with the conventional Fourier-basis. Our results show that the proposed method is more efficient than the conventional Fourier-based features in that they can obtain a higher speaker identification rate.

능동소나를 이용한 대잠전 환경에서 클러터는 표적탐지 및 추적성능을 저하시키는 가장 큰 원인 중 하나이다. 본 논문에서는 중주파수 능동소나에서 표적 피크 신호의 통계적 특징을 이용한 단일 핑 클러터 제거 기법을 제안한다. 기존의 표적 피크 영역을 제외한 잔향 존재 영역에서 오탐지율을 줄이는 기법이나 여러 핑을 누적하여 기동 패턴을 분석하여 표적과 클러터를 구분하는 기법들의 단점을 보완하기 위하여 단일 핑 데이터의 표적 피크 영역에서 통계적 특징 정보를 이용하여 클러터와 표적신호를 구분한다. 실제 표적을 이용한 해상실험에서 성능을 검증하였으며 기존 대비 클러터가 약 80 % 이상 제거되는 것을 확인하였다.

In active sonar system, clutters degrade performance of target detection/tracking and overwhelm sonar operators in ASW (Antisubmarine Warfare). Conventional clutter reduction algorithms using consistency of local peaks are studied in multi-ping data and tracking filter research for active sonar was conducted. However these algorithms cannot classify target and clutters in single ping data. This paper suggests a single ping clutter reduction approach to reduce clutters in mid-frequency active sonar system using echo shape features. The algorithm performance test is conducted using real sea-trial data in heavy clutter density environment. It is confirmed that the number of clutters was reduced by about 80 % over the conventional algorithm while retaining the detection of target.

Monitoring technology of machining has a long history since unmanned machining was introduced. Despite the long history, many researchers have presented new approaches continuously in this area. Sound based machine fault diagnosis is the process consisting of detecting automatically the damages that affect the machines by analyzing the sounds they produce during their operating time. The collected sound is corrupted by the surrounding work environment. Therefore, the most important part of the diagnosis is to find hidden elements inside the data that can represent the error pattern. This paper presents a feature extraction methodology that combines various digital signal processing and pattern recognition methods for the analysis of the sounds produced by tools. The magnitude spectrum of the sound is extracted using the Fourier analysis and the band-pass filter is applied to further characterize the data. Statistical functions are also used as input to the nonlinear classifier for the final response. The results prove that the proposed feature extraction method accurately captures the hidden patterns of the sound generated by the tool, unlike the conventional features. Therefore, it is shown that the proposed method can be applied to a sound based automatic diagnosis system.

본 논문에서는 자연 이미지가 갖는 통계적 일관성과 안개를 인식하는 시지각적 통계 특성을 이용하여 단일 안개 영상에서, 안개가 없는 참조 영상과의 비교 없이, 시지각적으로 안개 영상의 가시성을 예측한다. 제안하는 모델은 기존 안개 영상의 가시성 예측 방법들이 불가피하게 사용했던 추가 정보들, 예를 들면, 다수의 다양한 안개 영상, 차량 탑재 카메라의 지리적 위치 정보, 사람의 가시성 평가에 대한 학습 결과, 도로 선 혹은 교통 신호와 같이 안개 영상의 돋보이는 특정 물체 정보 등을 사용하지 않는다. 본 논문의 모델은 오직 테스트 안개 영상이 자연 현상에 의한 안개 영상 혹은 안개가 전혀 없는 영상에서 일관적으로 발견되는 통계적 특성으로부터 얼마나 떨어져 있는지 측정함으로써 안개 영상의 가시성을 예측한다. 시지각적으로 안개를 인식하여 일관된 통계를 나타내는 특징 인자들은 공간상의 자연 이미지 통계 모델과 안개 영상의 특징 (명암대비의 감소, 색상과 채도의 감소, 밝기의 증가)으로부터 유도된다. 제안하는 모델은 안개 영상의 전체 영역에 대한 가시성뿐만 아니라 각 관심 영역에서 패치 크기에 따른 지역적 안개 영상의 가시성도 예측할 수 있다. 본 모델의 성능분석을 위하여 사람이 직접 인지하는 가시성 측정 실험을 100 장의 다양한 안개 영상에 대해 수행하였다. 본 논문에서 제시한 모델을 통해 예측된 안개 영상의 가시성과 사람이 체감한 안개 영상의 가시성을 비교한 결과, 둘 사이에 매우 높은 상관관계가 있는 것으로 평가되었다. 본 논문이 제안하는 무참조 안개 영상의 가시성 예측 모델은 사람의 시지각적 특성을 활용한 새로운 방법으로, 향후 안개 영상의 가시성 향상 알고리듬 개발과 선 개발된 안개 제거 및 가시성 향상 알고리듬들의 성능을 정확히 평가할 수 있는 새로운 측정방법 개발 등에 매우 유용할 것으로 기대된다.

We propose a no-reference perceptual fog density and visibility prediction model in a single foggy scene based on natural scene statistics (NSS) and perceptual "fog aware" statistical features. Unlike previous studies, the proposed model predicts fog density without multiple foggy images, without salient objects in a scene including lane markings or traffic signs, without supplementary geographical information using an onboard camera, and without training on human-rated judgments. The proposed fog density and visibility predictor makes use of only measurable deviations from statistical regularities observed in natural foggy and fog-free images. Perceptual "fog aware" statistical features are derived from a corpus of natural foggy and fog-free images by using a spatial NSS model and observed fog characteristics including low contrast, faint color, and shifted luminance. The proposed model not only predicts perceptual fog density for the entire image but also provides local fog density for each patch size. To evaluate the performance of the proposed model against human judgments regarding fog visibility, we executed a human subjective study using a variety of 100 foggy images. Results show that the predicted fog density of the model correlates well with human judgments. The proposed model is a new fog density assessment work based on human visual perceptions. We hope that the proposed model will provide fertile ground for future research not only to enhance the visibility of foggy scenes but also to accurately evaluate the performance of defog algorithms.

Linguistics features are the information on the narrativestructure of a novel (Ehrlich 1996, Thompson and Hunston 2000,Fludernik 2009). Our research focuses on decisive linguistic factors forthe point of view in narratology. We extract statistics of 46 linguisticfeatures from 43 novels (2 million ejeols (combination of multiplemorphemes)) from the Sejong Corpus. Our statistical methodologyincludes dimensionality reduction, clustering, SVM and logisticregression. Our outcome explains the narratology of point of view. 1stpoint of view focuses on the speaker oriented narration. Meanwhile, 3rdpoint of view focuses on the narrator oriented narration. The usage ofpersonal pronouns is statistically biased between 1st and 3rd point ofview novels.

본 연구는 부산대학교 연구팀이 2013년 12월부터 약 2개월간 한국만화영상원의 의뢰를 받아 2000년 초부터 2013년까지 원고료를 지급받고 공식매체에서 연재된 한국 웹툰을 전수 조사한 '한국 웹툰DB 및 흐름 정리 연구' 자료를 토대로 통계 분석한 자료이다. 웹에 연재되는 만화를 의미하는 Webtoon은 전통적인 출판만화의 쇠퇴와 사회적 환경의 변화로 2000년대 이후 한국만화의 대표적 형식이자 주류 산업으로 발전해 왔으며, 현재 한국을 대표하는 문화콘텐츠로 불리고 있다. 본 조사연구는 웹툰이 시작된 2000년대 초부터 2014년 1월까지 총 13여 년간에 걸친 우리나라 웹툰 중에서 원고료를 받고 매체에 정식연재가 된 웹툰을 대상으로 수집, 정리되었다. 이 데이터를 토대로 작가, 작품의 수와 매체별 분포도, 장르와 연재 주기 등 전반적인 웹툰의 특성을 분석했다. 데이터 분석과 통계작업을 통해 살펴본 한국의 웹툰은 주요 포털의 연재 비중이 아직 높지만 서서히 플랫폼의 다변화가 진행되고 있으며, 작품의 연재 주기는 갈수록 짧아지는 경향을 보이고 있다. 장르적 특성으로는 드라마, 개그, 판타지, 액션 등의 만화의 전통적 인기장르는 여전히 건재하며 최근 사회적 트렌드에 맞게 역사물, 스포츠, 요리 등의 분야가 증가추세에 있다. 웹툰의 활용도 면으로는 릴레이 웹툰, 브랜드 웹툰과 같은 이벤트와 PPL식 상업성을 표방한 새로운 형태의 웹툰도 등장하고 있다. 이와 같은 현상은 작가와 매체, 발주자의 공동이익이 실현되는 한편, 나아가 웹툰의 가능성을 다양하게 실험하는 시도라고 할 수 있다. 그리고 웹툰의 저변확대가 활발해 지면서 성인물의 증가세도 눈여겨 볼만 하다. 본 연구대상은 무료웹툰을 제외한 원고료를 받는 작품을 기준으로 했으나 온라인 사이트의 속성 상 폐쇄되거나 유실된 작가와 작품을 모두 수집하지 못한 한계를 갖고 있으며, 무료 웹툰을 총망라한 전수조사가 앞으로 필요하다 하겠다. 그럼에도 한국 웹툰을 최초로 정식매체와 작품, 작가, 장르를 분류 조사하고 이를 토대로 웹툰의 현재를 가늠해 보는 기초자료로서의 의미를 찾고자 한다. 이 연구를 바탕으로 후속 웹툰연구가 활성화되고 보완되는 자료들이 한국의 만화산업과 학문적인 자료로서 활용되길 기대한다.

This study that had been conducted two months by a research team of Pusan National University at the request of Korea Manwha Contents Agency in Dec. 2013 is about the statistical analysis on 'Korean Webtoon DB and its Flow Report' which resulted from the complete survey of Korean webtoons which had been published with payment in official media from early 2000 to 2013. Webtoon which means the cartoons published on web has become a typical type of Korean cartoons and has developed into a main industry since 2000s when traditional published cartoons had declined and social environments had changed. Today, it represents cultural contents in Korea. This study collected the webtoons officially published in media with payment, among Korean webtoons having been published from the early 2000s to Jan. Based on the collected data, it analyzed the general characteristics of webtoons, including cartoonists, the number of cartoons, distribution chart of each media, genre, and publication cycle. According to the data analysis and statistics, a great deal of Korean webtoons are still published in main portal websites, but their platform is being diversified and a webtoon's publication cycle tends to be shortened. In terms of genre, traditional popular genres, such as drama, comic, fantasy, and action, are still popular, and the genres of history, sports, and food are on the rise along with a social trend. Regarding webtoon application, such events as relay webtoon and brand webtoon, and a new type of webtoon featuring PPL commercialism appear. Such phenomena can realize the common profits of cartoonists, media, and ordering bodies, and are various trials to test the possibility of webtoons. In addition, what needs to pay attention on in the expansion of webtoons is increasing webtoons for adults. The study subjects are the webtoons published with payment, excluding free webtoons. However, this study failed to collect the webtoons published on the online websites already closed, and the lost information on cartoonists and their lost webtoons, and it is necessary to conduct a complete survey on all webtoons including free ones. Despite the limitations, this study is meaningful in the points that it categorized and analyzed Korean webtoons accoridng to official media, webtoons, cartoonists, and genres and that it provided a fundamental material to understand the current conditions of webtoons. It is expected that this study will be able to contribute to activating more research on webtoons and producing more supplementary data which will be used for the Korean cartoon industry and academia.

This study employs statistical linguistic methods to conduct a fine-grained analysis of the combinatorial characteristics of the Korean particle ‘wa’ (와) in its two primary grammatical roles: the adverbial case particle and the conjunctive particle. Based on a large-scale written corpus, the research first conducts a frequency analysis of the morpheme categories that precede and follow ‘wa’. To move beyond raw frequency, the study then applies Mutual Information (MI) to quantitatively measure the strength of association between ‘wa’ and its adjacent morphemes. Furthermore, the Chi-square test is utilized to evaluate the statistical significance of these collocational patterns, distinguishing meaningful linguistic preferences from random co-occurrence. The analysis systematically compares the combinatorial profiles of the two functions based on these quantitative measures.

.

过去，在方言分类上，很少考虑到方言间的语法差异。本研究考察了10大方言区36个语言点的语义、词汇和语法差异，并把它反映在方言区的分类上。我们还认为，以往研究的问题在于观察并讨论的都是差异最大的、最极端的情形，而很少谈到两个极端现象中间存在的各种变体。我们全面考察了所有的变体，并对此设定了距离数值。本研究利用聚类分析和多维标度计算（MDS）方法，重新解释在汉语方言分区中还需要进一步解决的问题，如客家话和赣方言的分合问题、晉语的归属问题。