Earticle

본 연구는 유화 작품의 보존 상태를 모니터링하기 위하여 특징점 검출 알고리즘을 적용하여 작 품의 손상 여부를 살펴보고자 하였다. 미술품 보존 상태 모니터링은 주로 전문가의 육안 검사와 주기 적인 사진 촬영에 의존하고 있어, 시간과 비용이 많이 소요되며 객관적인 모니터링에 한계가 있다. 이 러한 문제를 해결하기 위해 본 연구에서는 SIFT, ORB, AKAZE, BRISK 등 다양한 특징점 검출 알 고리즘을 활용하였으며, 6가지 손상 유형(균열, 황변, 구김, 박리와 박락, 퇴색과 감변, 평면성 상실)이 포함된 유화 데이터셋에 대한 손상 검출 성능을 분석하였다. 검출률, 처리 시간, 새로운 특징점 검출 률, 메모리 사용량을 기준으로 평가한 결과, ORB 알고리즘이 가장 높은 검출률과 빠른 처리 속도를 보여 실시간 모니터링에 적합함을 확인하였다. 이를 통해 다량의 미술품을 보관하고 있는 미술관, 박 물관에서 미술품의 모니터링에 따른 알고리즘의 선택적 적용이 미술품 보존 관리에서의 손상 검출 효 율성을 향상시킬 수 있음을 확인하였다. 이러한 연구 결과는 미술관과 박물관에서 실제 작품 보존 관 리 시스템을 구축할 때 효과적인 알고리즘 선택의 기준을 제시할 수 있을 것으로 기대된다.

This study aimed to examine the condition of oil paintings by applying feature point detection algorithms to monitor their preservation status and assess potential damages. Over time, artworks are subject to natural deterioration and damage from environmental factors. Current conservation methods mainly rely on expert visual inspections and periodic photography, which are time-consuming, costly, and lack objectivity. To address these challenges, this study evaluates the performance of feature detection algorithms (SIFT, ORB, AKAZE, and BRISK) on a dataset of oil paintings with six types of damage: cracks, yellowing, wrinkles, delamination and flaking, fading, and loss of flatness. The algorithms were assessed based on detection rate, processing time, new feature detection rate, and memory usage. ORB demonstrated the highest detection rate and fastest processing speed, making it ideal for real-time monitoring. The results suggest that selective application of these algorithms in art galleries and museums could enhance the efficiency of damage detection, offering useful criteria for selecting effective algorithms in realworld art conservation management systems.

Defect detection is one of the key factors in fabric quality control. To improve the speed and accuracy of denim fabric defect detection, this paper proposes a defect detection algorithm based on cascading feature extraction architecture. Firstly, this paper extracts these weight parameters of the pre-trained VGG16 model on the large dataset ImageNet and uses its portability to train the defect detection classifier and the defect recognition classifier respectively. Secondly, retraining and adjusting partial weight parameters of the convolution layer were retrained and adjusted from of these two training models on the high-definition fabric defect dataset. The last step is merging these two models to get the defect detection algorithm based on cascading architecture. Then there are two comparative experiments between this improved defect detection algorithm and other feature extraction methods, such as VGG16, ResNet-50, and Xception. The results of experiments show that the defect detection accuracy of this defect detection algorithm can reach 94.3% and the speed is also increased by 1-3 percentage points.

이 연구는 유화에서 손상된 부위를 식별하고 추출하는 데 있어 SIFT, ORB, AKAZE, BRIST의 네 가지 특징 감지 알고리즘의 효능을 비교합니다. 다양한 열화 현상을 특징으로 하는 600개의 유화 이미지 데이터 세트를 컴파일하고 전처리했습 니다. 성능 평가 지표에는 감지된 특징의 수, 일치 능력, 일치 비율, 일관성 및 실행 시간이 포함되었습니다. 그 결과 BRIXE는 특징 감지량과 일치 수에서 우수하지만 처리 시간이 증가하는 대가를 치렀습니다. AKAZE와 SIFT는 각각 25.3%와 25%로 가장 높은 일치 비율을 보여 고품질 특징 감지를 시사했습니다. ORB는 이미지 전 반에서 가장 일관된 성능을 보였으며 절대 일치 수에서 2위를 차지했습니다. 이 연 구에 따르면 알고리즘 선택은 특정 작업 요구 사항을 기반으로 해야 합니다: 철저 한 특징 감지를 위한 BRIST, 고품질 일치를 위한 AKAZE 또는 SIFT, 일관되고 효율적인 성능을 위한 ORB. 이 연구는 유화의 손상 평가를 위한 객관적인 방법을 제공하고 잠재적으로 복원 프로세스를 간소화하며 보존 노력의 자원 할당을 최적화 함으로써 예술 보존에 기여합니다.

This study compares the efficacy of four feature detection algorithms: SIFT, ORB, AKAZE, and BRIST in identifying and extracting damaged areas in emulsions. A dataset of 600 emulsified images featuring various deterioration phenomena was compiled and preprocessed. Performance evaluation metrics included the number of detected features, matching ability, matching rate, consistency, and execution time. The results showed that BRIXE was excellent in feature detection and matching number, but at the cost of increasing processing time. AKAZE and SIFT showed the highest matching rates, 25.3% and 25%, respectively, suggesting high-quality feature detection. ORB performed most consistently across images and ranked second in absolute matching number. According to this study, algorithm selection should be based on specific task requirements: BRIST for thorough feature detection, AKAZE or SIFT for high-quality matching, and ORB for consistent and efficient performance. This study contributes to art conservation by providing an objective method for the evaluation of damage in emulsions, potentially simplifying the restoration process, and optimizing the resource allocation of conservation efforts.

본 논문에서는 효율적인 차량 객체를 추적하는 특징점 매칭을 이용한 다중 차량 객체 검 출 알고리즘을 제안한다. 제안하는 알고리즘은 효율적인 차량 객체 추적을 위해 FAST 알고 리즘을 이용해서 차량의 특징점을 추출한다. 그리고 5X5 영역으로 분할 된 영상에서 특징점 이 포함되면 True 포함되지 않으면 False로 해당 영역을 검은색으로 후처리하여 차량 객체을 제외한 불필요한 객체 정보를 제거한다. 그리고 후처리 된 영역을 차량의 최대 탐색창 크기 로 설정하고, 차량의 최외각 특징점을 이용한 최소 탐색창을 설정하여 Mean-Shift 알고리즘 의 탐색창 크기에 대한 단점을 보완하여 차량 객체 추적을 한다. 제안한 방법의 성능 평가하 기위해 SIFT, SURF 알고리즘을 비교하여 실험한다. 그 결과 SIFT 알고리즘에 비해서 약 4배 빠르고 SUFR 알고리즘의 처리 과정 보다는 효율적으로 검출하는 장점이 있다

In this paper, we propose a multi-vehicle object detection algorithm using feature point matching that tracks efficient vehicle objects. The proposed algorithm extracts the feature points of the vehicle using the FAST algorithm for efficient vehicle object tracking. And True if the feature points are included in the image segmented into the 5X5 region. If the feature point is not included, it is processed as False and the corresponding area is blacked to remove unnecessary object information excluding the vehicle object. Then, the post processed area is set as the maximum search window size of the vehicle. And A minimum search window using the outermost feature points of the vehicle is set. By using the set search window, we compensate the disadvantages of the search window size of mean-shift algorithm and track vehicle object. In order to evaluate the performance of the proposed method, SIFT and SURF algorithms are compared and tested. The result is about four times faster than the SIFT algorithm. And it has the advantage of detecting more efficiently than the process of SUFR algorithm.

본 논문은 객체 특징 탐색을 이용한 실시간 가상 모델 동기화 알고리즘을 제안한다. 제안된 알고리즘은 2차원 영상 에서 객체 특징 탐색을 사용하여, 실제 객체와 가상모델을 실시간으로 동기화 한다. 객체 특징 탐색 알고리즘은 색상을 이용 하여 객체를 개별적으로 분류하는 알고리즘과 각도를 이용하여 객체가 설치된 방향을 분석하는 알고리즘으로 구성된다. 제 안한 알고리즘을 사용하여 실제 물체의 움직임을 가상 모델에 동기화함으로써, 별도의 사용자 조작 도구 사용 없이 손을 사용하여 가상 물체를 움직이는 환경을 제공할 수 있다. 본 알고리즘은 향후 불특정한 모양과 색상 및 방향을 갖는 객체를 동기화하는 연구를 진행하며, VR/AR기술에 적용하여 사실적인 가상환경을 제공하는 것을 목표로 한다.

In this paper, we propose a real-time virtual model synchronization algorithm using object feature detection. The proposed algorithm may be useful to synchronize between real objects and their corresponding virtual models through object feature search in two-dimensional images. It consists of an algorithm to classify objects with colors individually, and an algorithm to analyze the orientation of objects with angles. We can synchronize the motion of the real object with the virtual model by providing the environment of moving the virtual object through the hand without specific controllers. The future research will include the algorithm to synchronize real object with unspecified shapes, colors, and directions to the corresponding virtual object.

메타학습 알고리즘인 Adaboost는 단순한 분류기(classifier)인 Haar-like feature 필터와 결 합하여 빠른 속도와 높은 정확도를 나타내기 때문에 많은 객체 검출에 분야에 사용되고 있으며 그 중 특히 얼굴 인식 분야에서 뛰어난 성 능을 나타내고 있다. 그러나 Adaboost 알고리즘은 얼굴인식 시 정면 얼굴과 같은 단일 모델 검출에는 매우 효율적 이지만 측면 얼굴과 정면 얼굴이 혼합된 모델에서의 검출 시 성능이 저하되는 문제점을 가 지고 있다. 이 문제를 해 결하기 위해 본 논문에서는 혼합모델의 검출 성능을 높이기 위해 인스턴스 정보를 저장하여 학습하는 알고리즘인 IBL(Instance Based Learning)를 Haar-like feature 필터와 결합한 알고리즘을 제안한다.

Adaboost is a meta learning algorithm and it is used many area of object detection. In the field of face detection it is used in conjunction with Haar-like feature filtering because of its fast speed and high accuracy. Adaboost shows very effective performance for face detection in single distribution model such as front face but it shows reduced performance on multiple distribution models such as mixture of front face and side face. As a result of reduced performance this paper proposed to improve performance of multiple distribution models using IBL (Instance Based Learning) algorithm combined with the Haar-like feature.

데이터 전처리 기법 중 하나인 특징 선택은 대규모 데이터셋을 다루는 다양한 응용분야에서 주요 연구 분야 중 하나로 각광받고 있다. 특징 선택은 패턴 인식, 기계학습 및 데이터 마이닝에서 사용됐고, 최근에는 텍스트 분류, 이미지 검색, 침입 탐지 및 게놈 분석과 같은 다양한 분야에 널리 적용되고 있다. 제안 방법은 메타 휴리스틱 알고리즘 중의 하나인 유전 알고리즘을 기반으로 한다. 특징 부분 집합을 찾는 방법은 크게 필터(filter) 방법과 래퍼(wrapper) 방법이 있는데, 본 연구에서는 최적의 특징 부분 집합을 찾기 위해 실제 분류기를 사용한 평가를 하는 래퍼 방법을 사용한다. 실험에 사용한 훈련 데이터셋은 클래스 불균형이 심하여 희소클래스에 대한 분류 성능을 높이기 어렵다. SMOTE 기법을 적용한 훈련 데이터셋을 사용하여 특징 선택을 하고 다양한 기계학습 알고리즘을 사용하여 선택한 특징들의 성능을 평가한다.

Feature selection, one of data preprocessing techniques, is one of major research areas in many applications dealing with large dataset. It has been used in pattern recognition, machine learning and data mining, and is now widely applied in a variety of fields such as text classification, image retrieval, intrusion detection and genome analysis. The proposed method is based on a genetic algorithm which is one of meta-heuristic algorithms. There are two methods of finding feature subsets: a filter method and a wrapper method. In this study, we use a wrapper method, which evaluates feature subsets using a real classifier, to find an optimal feature subset. The training dataset used in the experiment has a severe class imbalance and it is difficult to improve classification performance for rare classes. After preprocessing the training dataset with SMOTE, we select features and evaluate them with various machine learning algorithms.

Stable local feature and representation is a fundamental component of many image registration, 3D reconstruction and object recognition algorithms. SIFT is a good descriptor that encodes the salient aspects of the image gradient in the feature point’s neighborhood. This paper improved SIFT- based local image descriptor and proposed a SIFT feature matching algorithm based on improved 2DPCA which can eliminate both rows and columns of relevance. Experimental results show that improved 2DPCA-SIFT algorithm is relatively stable, accurate and fast.

Accurate detection performance of spectrum sensing will obviously affect the communication quality of licensed users and secondly users. Robust spectrum sensing algorithm can reduce interfere to primary users. This paper presents a new OFDM cyclic spectrum sensing scheme which based on overlay parts sub-carriers to original OFDM signals. In this paper, the parts sub-carriers come from original OFDM cyclic feature signals. Simulation and numerical results show that new method has a high success rate of sensing and a shorten detection time. Moreover, computational complexity is almost no change. Algorithm improved detection performance obviously and enhanced the effectiveness of anti-noise uncertainty and improved the robustness of detecting weak signal. Algorithm meets the spectrum sensing technology requirements of low SNR environment.

Iris biometric is most mature and secure authentication system as compared to other systems. Other authentication system does exist apart from biometrics such as PIN, password etc., which are not secure and more vulnerable to attacks and can be hacked or spoofed easily. Using, Iris biometric we can enhance overall performance in terms of accuracy, which is possible if and only if the pupil part must be removed perfectly and efficiently. Pupil part is unwanted part for our system. Pupil part is surrounded by iris part when extracted successfully we get two approximate concentric circle. Proposed scanning algorithm successfully extracts pupil part and defines iris part with less complexity and more efficiently. Iris part is consisting of patterns which are desired for authenticating a person and each patterns are represented in terms of feature vectors and stored in database. Proposed system focuses on feature extraction using five level decomposition technique implemented with haar, db2 and db4 and achieves high accuracy with reduced error rates. Due to reduced errors and considering lower half of iris part, proposed algorithm can be used for larger database such as for Aadhar because it takes less time for feature extraction and has less complexity with reduced mathematical burden on the system and improves good accuracy.

Feature selection algorithm plays a crucial role in intrusion detection, data mining and pattern recognition. According to some evaluation criteria, it gets optimal feature subset by deleting unrelated and redundant features of the original data set. Aiming at solving the problems about the low accuracy, the high false positive rate and the long detection time of the existing feature selection algorithm. In this paper, we come up with a feature selection algorithm towards efficient intrusion detection, this algorithm combines the correlation algorithm and redundancy algorithm to chooses the optimal feature subset. Experimental results show that the algorithm shows almost and even better than the traditional feature selection algorithm on the different classifiers.

A feature parameter modification algorithm is proposed to increase the accuracy of voice activity detection that based on support vector machine and energy acceleration parameters. The three energy acceleration parameters should have the equal importance for voice activity detection, in that all the three parameters can suitably characterize the classification features of speech and non-speech frames. For the three energy acceleration parameters, its minimum values vary a little; but its maximum values are greatly different. When radial basis function is chosen as the kernel function for voice activity detection, the coordinate values in Hilbert space is dominative determined by the energy values over a sub-region spectrum, the other two parameters only have a few contributions to it. The proposed algorithm extends the three energy acceleration parameters into the nearby or same order of magnitude. It make the three energy acceleration parameters at the same importance level in the calculation of coordinate values in Hilbert space, and can finally increase the accuracy of voice activity detection. The experimental results show that this algorithm can increase the accuracy of voice activity detection in the absence of noise and noise conditions.

Immature green citrus fruit detection using conventional color images is a challenging task due to fruit color similarity with the background, partial occlusion, varying illumination and shape irregularity. Therefore, most existing green fruits detection algorithms, which use color as the main discriminant feature, have a low recognition rate and a high rate of false positives. In this manuscript, we developed a novel Green Citrus fruit Detection algorithm based on the proposed Reticulate Grayladder Feature (GCDRGF), which contained 4 major steps: First, an 8-graylevel image was generated by the preprocessing steps of median filtering, histogram-based equalization and 8-graylevel discretization of the input raw image. Secondly, reticulate grayladders were obtained by a multidirectional scanning on the 8-graylevel image, and rule-based pseudo-grayladder removal strategies were used to remove false positives of target grayladders. Thirdly, grayladder clustering and fruit location fitting were used to generate candidate regions for target fruits. Finally, majority voting was performed to determine the results of candidate regions based on the analysis of apparent features and recticulate grayladders within candidate regions. The experimental results proved the effectiveness of the proposed reticulate grayladder feature and the corresponding detection algorithm with respect to various illuminant and imaging conditions. Compared with the existed eigenfruit algorithm, our algorithm has a higher rate of successful recognition and a lower rate of false positives, which helps to greatly improve the productivity of robotic operations.

This paper presents a Uyghur text-line separation algorithm based on classified connected components of Uyghur Handwritten scripts. In order to get the location of main text-lines, this paper proposes an adaptive painting and thinning algorithm. To insure the efficiency of text-lines segmentation, the post-processing of text-line detection procedures are introduced. The experimental results show that the algorithm is strongly robust for segmentation of text-lines with having some skewness, touching and overlapping, and small strokes remained.

Fingerprint feature-based text copy detection can rapidly identify the plagiarism, but suffers from the excessive fingerprint density. To resolve the problem, we propose a fingerprint feature extraction algorithm based on the optimal decision, combined with Winnowing algorithm and optimal decision model, and it can extract fingerprint feature from the hash values in the sliding window. The experimental results demonstrate that our algorithm can reduce the fingerprint density when the windows’ fingerprint feature is too adjacent, and the selected fingerprints can represent the text feature on the premise of the accuracy of the text copy and the algorithm.

This paper presents a novel algorithm for nighttime detection of the lane markers painted on a road at night. First of all, the proposed algorithm uses neighborhood average filtering, 8-directional Sobel operator and thresholding segmentation based on OTSU's to handle raw lane images taken from a digital CCD camera. Secondly, combining intensity map and gradient map, we analyze the distribution features of pixels on boundaries of lanes in the nighttime and construct 4 feature sets for these points, which are helpful to supply with sufficient data related to lane boundaries to detect lane markers much more robustly. Then, the searching method in multiple directions- horizontal, vertical and diagonal directions, is conducted to eliminate the noise points on lane boundaries. Adapted Hough transformation is utilized to obtain the feature parameters related to the lane edge. The proposed algorithm can not only significantly improve detection performance for the lane marker, but it requires less computational power. Finally, the algorithm is proved to be reliable and robust in lane detection in a nighttime scenario.

Human face avatar is important information in nowadays, such as describing real people in virtual world. In this paper, we have presented a face avatar creation and warping algorithm by using face feature analysis method, in order to detect face feature, we utilized local mean method based on facial feature appearance and face geometric information. Then detect facial candidates by using it's character in $YC_bC_r$ color space. Meanwhile, we also defined the rules which are based on face geometric information to limit searching range. For analyzing face feature, we used face feature points to describe their feature, and analyzed geometry relationship of these feature points to create the face avatar. Then we have carried out simulation on PC and embed mobile device such as PDA and mobile phone to evaluate efficiency of the proposed algorithm. From the simulation results, we can confirm that our proposed algorithm will have an outstanding performance and it's execution speed can also be acceptable.

In a cloud environment, performance degradation, or even downtime, of virtual machines (VMs) usually appears gradually along with anomalous states of VMs. To better characterize the state of a VM, all possible performance metrics are collected. For such high-dimensional datasets, this article proposes a feature extraction algorithm based on unsupervised fuzzy linear discriminant analysis with kernel (UFKLDA). By introducing the kernel method, UFKLDA can not only effectively deal with non-Gaussian datasets but also implement nonlinear feature extraction. Two sets of experiments were undertaken. In discriminability experiments, this article introduces quantitative criteria to measure discriminability among all classes of samples. The results show that UFKLDA improves discriminability compared with other popular feature extraction algorithms. In detection accuracy experiments, this article computes accuracy measures of an anomaly detection algorithm (i.e., C-SVM) on the original performance metrics and extracted features. The results show that anomaly detection with features extracted by UFKLDA improves the accuracy of detection in terms of sensitivity and specificity.

Purpose: This study investigated different band selection methods to classify spectrally similar data - obtained from aerial images of healthy citrus canopies and citrus greening disease (Huanglongbing or HLB) infected canopies - using small differences without unmixing endmember components and therefore without the need for an endmember library. However, large number of hyperspectral bands has high redundancy which had to be reduced through band selection. The objective, therefore, was to first select the best set of bands and then detect citrus Huanglongbing infected canopies using these bands in aerial hyperspectral images. Methods: The forward feature selection algorithm (FFSA) was chosen for band selection. The selected bands were used for identifying HLB infected pixels using various classifiers such as K nearest neighbor (KNN), support vector machine (SVM), naive Bayesian classifier (NBC), and generalized local discriminant bases (LDB). All bands were also utilized to compare results. Results: It was determined that a few well-chosen bands yielded much better results than when all bands were chosen, and brought the classification results on par with standard hyperspectral classification techniques such as spectral angle mapper (SAM) and mixture tuned matched filtering (MTMF). Median detection accuracies ranged from 66-80%, which showed great potential toward rapid detection of the disease. Conclusions: Among the methods investigated, a support vector machine classifier combined with the forward feature selection algorithm yielded the best results.

장면 전환 검출에서 단일 특성을 사용하는 경우 발생 가능한 정확도 감소의 문제를 해결하기 위해 색상 히스토그램 분포 차 분석과 특징점 추출 알고리즘을 활용한 방법을 제안한다.