Deep Convolution Neural Networks 이용하여 결함 검출을 위한 결함이 있는 철도선로표면 디지털영상 재 생성

김현호 ( Hyeonho Kim ),한석민 ( Seokmin Han ) 한국인터넷정보학회 2020 인터넷정보학회논문지 Vol.21 No.6

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[14] 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 영상처리 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[15, 16]. 본연구는 이러한 기계적인 영상획득 장치의 운용비용을 감소시키고 보완하기 위해 대표적인 영상생성관련 딥러닝 모델인 생성적 적대적 네트워크[1]의 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 모델[2]에 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 원본결함지도(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. This study was carried out to generate various images of railroad surfaces with random defects as training data to be better at the detection of defects. Defects on the surface of railroads are caused by various factors such as friction between track binding devices and adjacent tracks and can cause accidents such as broken rails [14], so railroad maintenance for defects is necessary. Therefore, various researches on defect detection and inspection using image processing or machine learning on railway surface images have been conducted to automate railroad inspection and to reduce railroad maintenance costs. In general, the performance of the image processing analysis method and machine learning technology is affected by the quantity and quality of data. For this reason, some researches require specific devices or vehicles to acquire images of the track surface at regular intervals to obtain a database of various railway surface images [15, 16]. On the contrary, in this study, in order to reduce and improve the operating cost of image acquisition, we constructed the 'Defective Railroad Surface Regeneration Model' by applying the methods presented in the related studies of the Generative Adversarial Network (GAN) [1]. Thus, we aimed to detect defects on railroad surface even without a dedicated database. This constructed model is designed to learn to generate the railroad surface combining the different railroad surface textures and the original surface, considering the ground truth of the railroad defects. The generated images of the railroad surface were used as training data in defect detection network [2], which is based on Fully Convolutional Network (FCN) [3]. To validate its performance, we clustered and divided the railroad data into three subsets, one subset as original railroad texture images and the remaining two subsets as another railroad surface texture images. In the first experiment, we used only original texture images for training sets in the defect detection model. And in the second experiment, we trained the generated images that were generated by combining the original images with a few railroad textures of the other images. Each defect detection model was evaluated in terms of 'intersection of union(IoU)' and F1-score measures with ground truths. As a result, the scores increased by about 10~15% when the generated images were used, compared to the case that only the original images were used. This proves that it is possible to detect defects by using the existing data and a few different texture images, even for the railroad surface images in which dedicated training database is not constructed.

지진시 적재설비의 내진장치 기술 및 낙하방지 장치 기초기술 개발

김권영(Gwon-Young Kim),최재성(Jae-Sung Choi),김창열(Chang-Yeol Kim),김정회(Jeong-Hoe Kim),허광희(Gwang-Hee Heo),최형석(Hyoung-Suk Choi) 한국소음진동공학회 2017 한국소음진동공학회 학술대회논문집 Vol.2017 No.10

지진시 소화배관 기능 유지를 위한 흔들림방지버팀대 성능인정 시험

김권영(Gwon-Young Kim),최재성(Jae-Sung Choi),김창열(Chang-Yeol Kim),김정회(Jeong-Hoe Kim),이규형(Kyu-Hyeong Lee),윤대진(Dae-Jin Yun) 한국소음진동공학회 2018 한국소음진동공학회 학술대회논문집 Vol.2018 No.4

지진대비 물류센터 적재설비의 내진성능 향상을 위한 제품개발 및 현장적용

김권영(Gwon-Young Kim),최재성(Jae-Sung Choi),김창열(Chang-Yeol Kim),김정회(Jung-Hole Kim),이규형(Kyu-Hyoung Lee),윤대진(Dae-Jin Yun) 한국소음진동공학회 2018 한국소음진동공학회 학술대회논문집 Vol.2018 No.10

성인여성의 척추측만증과 하지 근력비율 및 통증과의 관계 연구

김권영(Kim, Kwon-Young),홍완기(Hong, Wan-Ki),김재윤(Kim, Jae-Youn) 한국체육과학회 2018 한국체육과학회지 Vol.27 No.6

The purpose of this study was to investigate the mechanism of musculoskeletal disorders in middle-aged women by verifying the degree of scoliosis, lower extremity muscle strength, and knee pain. To do this, 35 middle-aged women were divided into scoliosis group (SG) and normal group (NG). Results of X-ray, isokinetic muscle function test and WOMAC test were analyzed. Respectively. First, knee pain score of scoliosis group was significantly higher than that of normal group. Second, there was a statistically significant correlation between the degree of cobb angle and the extension of knee strength. Third, there was a positive correlation between the degree of cobb angle and knee pain index (WOMAC). Based on the results, it was found that the scoliosis of middle-aged women had characteristics of lower extremity functional index and pain index, and the identification of these characteristics would require intervention of scoliosis in the design of exercise program for lower extremity musculoskeletal disorders.

가잠(家蠶)의 바이러스성(性) 연화병(軟化病)에 관한 연구 -장려품종(奬勵品種)에 대한 저항성(抵抗性) 검정(檢定)-

김권영 ( Keun Young Kim ),강석권 ( Seok Kwon Kang ),이재창 ( Jae Chang Lee ) 한국잠사학회 1978 한국잠사곤충학회지 Vol.20 No.2

현장려잠품종에 대하여 FV 및 이나주바이러스의 2령기잠 경구접종에 의한 품종간 저항성을 비교 검정한 결과는 다음과 같다. 1. FV에 대한 품종별 저항성은 한생 1호×한생 2호가 가장 강하였고 무등×선호가 가장 약했다. 2. 이나주 바이러스에 대한 저항성은 잠117×잠118, 경추×소일, 무등×선호, 한생 1호×한생 2호 및 한생 3호×4호 등 5개 품종은 완전저항성을 나타냈으며 잠115×잠116이 가장 약했다. The infectious flacherie viruses are serious diseases in the silkworm, which affects the cocoon crops. However, there is only a few study on this diseases up-to-date in Korea and, in this experiment, the authors investigated the resistance of the silkworm varieties to flacherie and Ina-flacherie viruses by peroral infection. A cross, Hansaeng #1×Harisaeng #2, showed the highest resistance to the flacherie virus a Mudeung×Geumho showed the lowest resistance among the examined varieties. It seemed that the varietal difference of resistance against Ina-flacherie virus was appeared and Jam117×Jam118, Gyeongchu×Yeonil, Mudeung×Geumho, Hansaeng#1×Hansaeng#2 and Hansaeng#3×Hansaeng#4 showed non susceptibility by peroral infection. However it was shown that Jam115×Jam116 had the lowest resistance to Ina-flacherie virus.

장기간 교대근무가 심혈관계 위험요인에 미치는 영향

김권영(Kwon Yung Kim),이경일(Kyung Il Lee),이철갑(Chul Gab Lee),홍완기(Wan Ki Hong) 한국사회체육학회 2007 한국사회체육학회지 Vol.0 No.30

To evaluate the relation between shift work term and risk factor for cardiovascular system this study was conducted at 126persons a city company. The subjects were divided by Shift work term into 3group, that were Non-Longer Work Group(NLWG: <10year), Short Work Group(SWG: ≥10year~<20year), Longer Work Group(LWG: ≥20year). Systolic blood pressure was higher than NLWG & SWG significantly in Causal risk factors, and there were no significant differences in other checkups. TG of NLWG & LWG shows high aspect significantly in conditional and interposition risk factors. and there were no significant differences in other checkups. The Age and "Glucose", "TG", "systolic blood pressure" shows significant static correlation in correlation analysis between subordinate checkups of dagerous factors. Besides, Glucose and TC shows significant correlation to each other and TC & LDL-C shows significant correlation to each other. Also, TG and HDL-C shows significant correlation to each other and BMI and systolic blood pressure shows significant correlation to each other.

성인여성의 하지 정렬과 통증, 신체적 기능과의 관계 : 하지근력의 조절효과

김권영(Kim, Kwon-Young),홍완기(Hong, Wan-Ki),김재윤(Kim, Jae-Youn) 한국체육과학회 2020 한국체육과학회지 Vol.29 No.2

This study was conducted to find out the relationship between lower limb alignment, pain and physical function of adult women and to see if muscle strength is involved in the coordination of these relationships. For 151 adult women, x-rays and muscle functional tests and WOMACs were used to calculate the pain and physical function index. The results were derived by analyzing the structure equation model method. First, the lower limb alignment angle had a relationship between the muscle strength and negative(-) and the pain and physical function had a positive(+) relationship. Second, The muscle strength of had no adjustment effect to relieve pain in relation to the lower limb alignment angle and pain. Third, Knee muscular strength had no adjustment effect to improve function in relation to lower limb alignment and physical function. Consequently, the lower limb alignment angle was related to pain and physical function, but the muscle strength was not involved in the adjustment effect in the relationship with them.

Identification of the Polymorphisms in IFITM2 and IFITM5 Genes and their Association with Ulcerative Colitis

Hun-Soo Kim(김헌수),Ji-Su Mo(모지수),Khondoker Jahengir Alam(알롬 콘도칼자항길),Won-Cheol Park(박원철),Keun Young Kim(김권영),Soo-Cheon Chae(채수천) 한국생명과학회 2015 생명과학회지 Vol.25 No.1

Interferon inducible transmembrane protein (IFITM) family 유전자는 인터페론(IFNs)의 동형 세포부착 기능 및 세포의 항-증식 활성과 같은 몇 가지 세포증식 과정에 연관되어 있다. 본 연구에서는 IFITM2 및 IFITM5 SNPs이 궤양성대장염의 감수성과 연관되어 있는지 알아 보고자 했다. 본 연구에서 직접 염기서열 분석법을 사용하여 IFITM2 유전자에서 총 13개, IFITM5 유전자에서는 12개의 유전적 변이를 발굴하였다. 이들의 SNPs의 유전자형 분석은 PCR-RFLP 법과 Taq-Man probe 분석법을 사용하였고, 일배체형 빈도 분석은 EM algorithm을 사용하여 분석하였다. 궤양성대장염 환자에서 IFITM2 및 IFITM5 SNPs의 유전자형과 대립유전자 빈도는 건강인 대조군과 비교했을 때 유의성이 없었다. 궤양성대장염 환자와 정상인 대조군에서 IFITM1의 rs77537847, IFITM2의 rs909097, IFITM5의 rs56069858을 지표로 하는 유전자형 조합 빈도를 분석한 결과 주된 유전자형 조합빈도에서는 유의성이 없는 것으로 나타났으나, 궤양성대장염 환자와 건강인 대조군의 GGT 유전자형조합 빈도 분석에서는 유의하게 다른 차이를 보였다(p=0.002). 이러한 결과에 의거하여 IFITMs의 SNPs 유전자형 조합이 궤양성대장염의 감수성과 연관성이 있고, 궤양성대장염의 유용한 유전자 마커로 사용 할 수 있다고 생각된다. Interferon inducible transmembrane protein (IFITM) family genes have been implicated in various cellular processes such as the homotypic cell adhesion functions of IFNs and cellular anti-proliferative activities. The present study aimed to investigate whether the polymorphisms of the IFITM2 and IFITM5 genes are associated with susceptibility to UC. We identified a total of thirteen polymorphisms (eleven SNPs and two variations) in the IFITM2 gene and twelve polymorphisms (eleven SNPs and one variation) in the IFITM5 gene, by the direct sequencing method. Genotype analysis in the IFITM2 and IFITM5 SNPs was performed by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP) and Taq-Man probe analysis, and the haplotype frequencies of IFITM2 and IFITM5 SNPs for multiple loci were estimated using the expectation maximization (EM) algorithm. The genotype and allele frequencies of IFITM2 SNPs, as well as IFITM5 SNPs, in UC patients were not significantly different from those of the healthy controls. We also analyzed the combined frequencies of rs77537847 of IFITM1, rs909097 of IFITM2, and rs56069858 of IFITM5 in the UC patients and the healthy controls. Although the distribution of the major combined genotype frequency did not differ significantly between the healthy controls and the UC patients, the GGT combined frequency in the healthy controls was significantly different from that in the UC patients (P=0.002). This result suggests that the combined genotype of the IFITMs polymorphisms may be associated with a susceptibility to UC and could be a useful genetic marker for UC.

적재설비의 내진장치 및 낙하방지 장치 개발을 위한 성능 시험

김창열(Chang-Yeol Kim),최재성(Jae-Sung Choi),김권영(Gwon-Young Kim),김정회(Jung-Hoe Kim),이규형(Kyu-Hyoung Lee),윤대진(Dae-Jin Yun) 한국소음진동공학회 2018 한국소음진동공학회 학술대회논문집 Vol.2018 No.4