• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제12권1호
• /
• pp.454-475
• /
• 2018

This paper presents a novel Android malware classification model planned to classify and categorize Android malicious code at Drebin dataset. The amount of malicious mobile application targeting Android based smartphones has increased rapidly. In this paper, Restricted Boltzmann Machine and Deep Belief Network are used to classify malware into families of Android application. A texture-fingerprint based approach is proposed to extract or detect the feature of malware content. A malware has a unique "image texture" in feature spatial relations. The method uses information on texture image extracted from malicious or benign code, which are mapped to uncompressed gray-scale according to the texture image-based approach. By studying and extracting the implicit features of the API call from a large number of training samples, we get the original dynamic activity features sets. In order to improve the accuracy of classification algorithm on the features selection, on the basis of which, it combines the implicit features of the texture image and API call in malicious code, to train Restricted Boltzmann Machine and Back Propagation. In an evaluation with different malware and benign samples, the experimental results suggest that the usability of this method---using Deep Belief Network to classify Android malware by their texture images and API calls, it detects more than 94% of the malware with few false alarms. Which is higher than shallow machine learning algorithm clearly.

• 한국인터넷방송통신학회논문지
• /
• 제20권4호
• /
• pp.201-205
• /
• 2020

구조적인 텍스처는 텍스처를 구성하는 기본요소인 텍셀 (texel)이 규칙적으로 반복되는 형태로 정의된다. 구조적 텍스처 분석/인식은 직물류의 자동검사, 금속표면 자동테스트 및 마이크로 이미지의 자동 분석 등, 산업적인 응용이 다양하다. 본 논문에서는 구조적 텍스처 분석을 위한 합성곱신경망 (Convolution Neural Network, CNN) 기반의 시스템을 제안한다. 제안한 방법은 합성곱신경망이 분류 대상 텍스처들의 구성 요소인 텍셀을 학습한다. 인식 단계에서는 입력되는 텍스처 영상에서 얻은 부분 영상을 이용하여 학습된 합성곱신경망이 텍스처를 인식하다. 실제 구현 및 실험을 통하여 제안된 방법의 우수성을 보인다.

• 한국컴퓨터정보학회논문지
• /
• 제21권11호
• /
• pp.17-21
• /
• 2016

Heterogeneity assessment of tumor in oncology is important for diagnosis of cancer and therapy. The aim of this study was performed assess heterogeneity tumor region in PET image using texture analysis. For assessment of heterogeneity tumor in PET image, we inserted sphere phantom in torso phantom. Cu-64 labeled radioisotope was administrated by 156.84 MBq in torso phantom. PET/CT image was acquired by PET/CT scanner (Discovery 710, GE Healthcare, Milwaukee, WI). The texture analysis of PET images was calculated using occurrence probability of gray level co-occurrence matrix. Energy and entropy is one of results of texture analysis. We performed the texture analysis in tumor, liver, and background. Assessment textural features of region-of-interest (ROI) in torso phantom used in-house software. We calculated the textural features of torso phantom in PET image using texture analysis. Calculated entropy in tumor, liver, and background were 5.322, 7.639, and 7.818. The further study will perform assessment of heterogeneity using clinical tumor PET image.

• 한국토양비료학회지
• /
• 제48권5호
• /
• pp.351-360
• /
• 2015

Human influence on soil formation has dramatically increased with human civilization and industry development. Increase of anthropogenic soils induced researches on the anthropogenic soils; classification, chemical and physical characteristics of anthropogenic soils and plant growth from anthropogenic soils. However there have been no comprehensive analyses on soil pore or physical properties of anthropogenic soils from 3 dimensional images in Korea. The objectives of this study were to characterize physical properties of anthropogenic paddy field soils by depth and to find differences between natural and anthropogenic paddy field soils. Soil samples were taken from two anthropogenic and natural paddy field soils; anthropogenic (A_c) and natural (N_c) paddy soils with topsoil of coarse texture and anthropogenic (A_f) and natural (N_f) paddy soils with topsoil of fine texture. The anthropogenic paddy fields were reestablished during the Arable Land Remodeling Project from 2011 to 2012 and continued rice farming after the project. Natural paddy fields had no artificial changes or disturbance in soil layers up to 1m depth. Samples were taken at three different depths and analyzed for routine physical properties (texture, bulk density, etc.) and pore properties with computer tomography (CT) scans. The CT scan provided 3 dimensional images at resolution of 0.01 mm to calculate pore radius size, length, and tortuosity of soil pores. Fractal and configuration entropy analyses were applied to quantify pore structure and analyze spatial distribution of pores within soil images. The results of measured physical properties showed no clear trend or significant differences across depths or sites from all samples, except the properties from topsoils. The results of pore morphology and spatial distribution analyses provided detailed information of pores affected by human influences. Pore length and size showed significant decrease in anthropogenic soils. Especially, pores of A_c had great decrease in length compared to N_c. Fractal and entropy analyses showed clear changes of pore distributions across sites. The topsoil layer of A_c showed more degradation of pore structure than that of N_c, while pores of A_f topsoil did not show significant degradation compared with those of N_f. These results concluded that anthropogenic soils with coarse texture may have more effects on pore properties than ones with fine texture. The reestablished paddy fields may need more fundamental remediation to improve physical conditions.

• 한국생활과학회지
• /
• 제20권5호
• /
• pp.1025-1034
• /
• 2011

The purpose of this study was to examine the evaluation of fabric characteristics on the drapability, texture image and preference of blouse fabrics, and to analyze the effects of the texture image, objective and subjective drapability on the preference. As specimen, silk and polyester fabrics were collected. 52 female subjects evaluated 16 specimens with semantic differential scale of 18 fabric image and 20 sensibility. Data were analyzed through factor analysis, pearson correlational coefficient using spss win 12.0. For the evaluation, structural characteristics such as fiber contents, weave type, weight and thickness were analyzed. The results were as follows: The evaluation results of objective and subjective drapability showed differences. Sensory image factors of blouse fabrics were 'surface smoothness', 'elasticity', 'weight' and 'flexibility'. Sensibility image factors were 'elegance', 'classic', 'characteristic' and 'mannish'. 'Elegance', 'classic' and 'characteristic' of sensibility images showed high correlation with 'surface smoothness' and 'elasticity' of sensory image, also 'mannish' of sensibility image showed significant correlation with 'weight' of sensory image. The significant fabric characteristics affecting objective drapability were density, weight, thickness. The significant texture image factors affecting objective drapability were 'weight', 'flexibility' of sensory image and 'elegance' of sensibility image. On the other hand, the significant factors affecting subjective drapability were thickness of fabric characteristics and 'elegance', 'characteristic', 'mannish' of sensibility images. 'Elegance', 'characteristic' and 'classic' of sensibility image, 'elasticity' of sensory image and subjective drapability affected on the purchase preference.

• 대한의용생체공학회:의공학회지
• /
• 제44권5호
• /
• pp.336-345
• /
• 2023

The present study was designed to evaluate changes in neuromuscular properties and the structural and qualitative characteristics of muscles during submaximal isometric contractions at low-to-relatively vigorous target forces and to determine their influence on force steadiness (FS). Thirteen young adult males performed submaximal isometric knee extensions at 10, 20, 50, and 70% of their maximal voluntary isometric contraction using their non-dominant legs. During submaximal contractions, we recorded force, EMG signals from vastus medialis (VM), vastus lateralis (VL), and rectus femoris (RF), and ultrasound images from the distal RF (dRF). Force and EMG standard deviation (SD) and coefficient of variation (CV) values were used to measure FS and EMG steadiness, respectively. Muscle thickness (MT), pennation angle (PA), echo intensity (EI), and texture features were calculated from ultrasound images to assess the structural and qualitative characteristics of the muscle. FS, neuromuscular properties, and texture features showed significant differences across different force levels. Additionally, there were significant differences in EMG_CV among the quadriceps at the 50% and 70% force levels. The results of correlation analysis revealed that FS had a significant relationship with EMG_CV in VM, VL, and RF, as well as with the texture features of dRF. This study's findings demonstrate that EMG steadiness and texture features are influenced by the magnitude of the target force and are closely related to FS, indicating their potential contribution to force output control.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 하계종합학술대회 논문집(4)
• /
• pp.11-14
• /
• 2000

We present an image retrieval method that improves retrieval rate by using the fusion of histogram and wavelet moment features. The key idea is that images similar to a query image are selected in DB by using the wavelet moment features. Then the result images are retrieved from the selected images by using histogram method. In order to evaluate the performance of the proposed method, we use Brodatz texture database, MPEG-7 T1 database and Corel Draw photo. Experimental result shows that the proposed method is better than each of histogram method and wavelet moment method.

• 인터넷정보학회논문지
• /
• 제21권6호
• /
• pp.23-31
• /
• 2020

본 연구는 철도표면상에 발생하는 노후 현상 중 하나인 결함 검출을 위해 학습데이터를 생성함으로써 결함 검출 모델에서 더 높은 점수를 얻기 위해 진행되었다. 철도표면에서 결함은 선로결속장치 및 선로와 차량의 마찰 등 다양한 원인에 의해 발생하고 선로 파손 등의 사고를 유발할 수 있기 때문에 결함에 대한 철도 유지관리가 필요 하다. 그래서 철도 유지관리의 자동화 및 비용절감을 위해 철도 표면 영상에 영상처리 또는 기계학습을 활용한 결함 검출 및 검사에 대한 다양한 연구가 진행되고 있다. 일반적으로 영상 처리 분석기법 및 기계학습 기술의 성능은 데이터의 수량과 품질에 의존한다. 그렇기 때문에 일부 연구는 일반적이고 다양한 철도표면영상의 데이터베이스를 확보하기위해 등간격으로 선로표면을 촬영하는 장치 또는 탑재된 차량이 필요로 하였다. 본연구는 이러한 기계적인 영상획득 장치의 운용비용을 감소시키고 보완하기 위해 대표적인 영상생성관련 딥러닝 모델인 생성적 적대적 네트워크의 기본 구성에서 여러 관련연구에서 제시된 방법을 응용, 결함이 있는 철도 표면 재생성모델을 구성하여, 전용 데이터베이스가 구축되지 않은 철도 표면 영상에 대해서도 결함 검출을 진행할 수 있도록 하였다. 구성한 모델은 상이한 철도 표면 텍스처들을 반영한 철도 표면 생성을 학습하고 여러 임의의 결함의 위치에 대한 Ground-Truth들을 만족하는 다양한 결함을 재 생성하도록 설계하였다. 재생성된 철도 표면의 영상들을 결함 검출 딥러닝 모델에 학습데이터로 사용한다. 재생성모델의 유효성을 검증하기 위해 철도표면데이터를 3가지의 하위집합으로 군집화 하여 하나의 집합세트를 원본 영상으로 정의하고, 다른 두개의 나머지 하위집합들의 몇가지의 선로표면영상을 텍스처 영상으로 사용하여 새로운 철도 표면 영상을 생성한다. 그리고 결함 검출 모델에서 학습데이터로 생성된 새로운 철도 표면 영상을 사용하였을 때와, 생성된 철도 표면 영상이 없는 원본 영상을 사용하였을 때를 나누어 검증한다. 앞서 분류했던 하위집합들 중에서 원본영상으로 사용된 집합세트를 제외한 두 개의 하위집합들은 각각의 환경에서 학습된 결함 검출 모델에서 검증하여 출력인 픽셀단위 분류지도 영상을 얻는다. 이 픽셀단위 분류지도영상들과 실제 결함의 위치에 대한 원본결함 지도(Ground-Truth)들의 IoU(Intersection over Union) 및 F1-score로 평가하여 성능을 계산하였다. 결과적으로 두개의 하위집합의 텍스처 영상을 이용한 재생성된 학습데이터를 학습한 결함 검출모델의 점수는 원본 영상만을 학습하였을 때의 점수보다 약 IoU 및 F1-score가 10~15% 증가하였다. 이는 전용 학습 데이터가 구축되지 않은 철도표면 영상에 대해서도 기존 데이터를 이용하여 결함 검출이 상당히 가능함을 증명하는 것이다.

• 한국콘텐츠학회논문지
• /
• 제14권11호
• /
• pp.362-369
• /
• 2014

본 연구는 골다공증 환자의 Digital 요추 측부 영상을 이용하여 질감특징의 통계적 분석으로 컴퓨터 보조진단 시스템 구현과 질병의 조기진단 및 치료를 위한 실험적인 모형 연구로 신뢰성 있는 보조적 진단 정보를 제공함으로써 골다공증에 대한 정확한 진단 방향을 제시하고자 하였다. 이를 위해서 정상인의 Digital 방사선 요추 측부 영상과 골다공증 환자의 Digital 방사선 요추 측부 영상을 실험 영상으로 하여 설정된 ROI에 대한 통계적 질감특징 값을 6가지 parameter로 나타냈다. 골다공증에 대한 질감특징분석 값 중 Average Gray Level에서 95%로 최고 높은 인식률을 나타내었고, Uniformity에서 80%로 가장 낮은 인식률을 나타내었다. 또한 Average Contrast에서 82.5%, Smoothness에서 90%, Skewness에서 87.5%, Entropy에서 87.5%를 나타내어 6가지 Parameter에서 모두 80%이상의 높은 인식률을 나타내 알고리즘의 안정성을 입증하였다. 따라서 본 연구 결과를 토대로 의료영상의 컴퓨터자동진단 시스템으로 발전된 프로그램을 coding 한다면 의료영상의 병소부위 자동검출, 질병 진단을 위한 예비 진단자료, 질병의 확진을 위한 자료제공, 제한된 장비로도 진단 가능, 의료영상의 판독시간 단축에 유용하게 사용될 수 있으리라 사료된다.

• The Journal of the Acoustical Society of Korea
• /
• 제15권2E호
• /
• pp.85-91
• /
• 1996

New very low bit rate segmentation video image coding technique is proposed by segmenting image into textually homogeneous regions. Regions are classified into one of three perceptually distinct texture classes(perceived constant intensity, smooth texture, and rough texture) using the Human Visual System(HVS) and the fractals. To design very low bit rate video image coder, it is very important to determine the best block size for estimation the fractal dimension and the thresholding of the fractal dimension for each texture class. Good quality reconstructed images are obtained with about 0.10 to 0.21 bit per pixel(bpp) for many different types of imagery.