http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
( Ohkyung Kwon ),( Hyung Gu Lee ),( Sang-Yun Yang ),( Hyunbin Kim ),( Se-Yeong Park ),( In-Gyu Choi ),( Hwanmyeong Yeo ) 한국목재공학회 2019 목재공학 Vol.47 No.3
In our previous study, the LeNet3 model successfully classified images from the transverse surfaces of five Korean softwood species (cedar, cypress, Korean pine, Korean red pine, and larch). However, a practical limitation exists in our system stemming from the nature of the training images obtained from the transverse plane of the wood species. In real-world applications, it is necessary to utilize images from the longitudinal surfaces of lumber. Thus, we improved our model by training it with images from the longitudinal and transverse surfaces of lumber. Because the longitudinal surface has complex but less distinguishable features than the transverse surface, the classification performance of the LeNet3 model decreases when we include images from the longitudinal surfaces of the five Korean softwood species. To remedy this situation, we adopt ensemble methods that can enhance the classification performance. Herein, we investigated the use of ensemble models from the LeNet and MiniVGGNet models to automatically classify the transverse and longitudinal surfaces of the five Korean softwoods. Experimentally, the best classification performance was achieved via an ensemble model comprising the LeNet2, LeNet3, and MiniVGGNet4 models trained using input images of 128 × 128 × 3 pixels via the averaging method. The ensemble model showed an F1 score greater than 0.98. The classification performance for the longitudinal surfaces of Korean pine and Korean red pine was significantly improved by the ensemble model compared to individual convolutional neural network models such as LeNet3.
Automatic Wood Species Identification of Korean Softwood Based on Convolutional Neural Networks
Ohkyung Kwon,Hyung Gu Lee,Mi-rim Lee,Sujin Jang,Sang-yun Yang,Se-yeong Park,In-gyu Choi,Hwanmyeong Yeo 한국목재공학회 2017 목재공학 Vol.45 No.6
Automatic wood species identification systems have enabled fast and accurate identification of wood species outside of specialized laboratories with well-trained experts on wood species identification. Conventional auto-matic wood species identification systems consist of two major parts: a feature extractor and a classifier. Feature extractors require hand-engineering to obtain optimal features to quantify the content of an image. A Convolutional Neural Network (CNN), which is one of the Deep Learning methods, trained for wood species can extract intrinsic feature representations and classify them correctly. It usually outperforms classifiers built on top of extracted features with a hand-tuning process. We developed an automatic wood species identification system utilizing CNN models such as LeNet, MiniVGGNet, and their variants. A smartphone camera was used for obtaining macroscopic images of rough sawn surfaces from cross sections of woods. Five Korean softwood species (cedar, cypress, Korean pine, Korean red pine, and larch) were under classification by the CNN models. The highest and most stable CNN model was LeNet3 that is two additional layers added to the original LeNet architecture. The accuracy of species identi-fication by LeNet3 architecture for the five Korean softwood species was 99.3%. The result showed the auto-matic wood species identification system is sufficiently fast and accurate as well as small to be deployed to a mobile device such as a smartphone.
Kwon, Ohkyung Korean Society of Microscopy 2014 Applied microscopy Vol.44 No.2
Layered structures of fiber cell wall of Korean red pine (Pinus densiflora) were investigated by confocal reflection microscopy (CRM). CRM micrographs revealed detailed structures of the fiber cell wall such as S1, S2, and S3 layers as well as transition layers (S12 and S23 layers), which are present between the S1, S2, and S3 layers. Microfibril angle (MFA) measurement was possible for the S2 and S3 layer in the cell wall. The experimental results suggest that CRM is a versatile microscopic method for investigation of layered structures and MFA measurement in individual sub layer of the tracheid cell wall.
하이브리드 방식을 적용한 흡수식 냉방시스템의 실험적 연구
권오경(Ohkyung Kwon),차동안(Dongan Cha),김효상(Hyosang Kim),우성민(Sungmin Woo) 대한기계학회 2010 대한기계학회 춘추학술대회 Vol.2010 No.11
In this paper, the performance evaluation for a solar/gas hybrid absorption system using LiBr-H₂O is experimentally studied. In order to use solar energy more effectively, a new type of solar/gas driving hybrid absorption system is designed. This system operates in solar energy without gas consumption when solar radiation is high and in gas energy when solar radiation is low. So the objective of this paper is to investigate the cooling characteristics of a hybrid absorption system using solar and gas energy. In this system, three different modes are conducted: single, double and hybrid operation mode. A prototype is designed and made in the present study. The results show that the COP in the hybrid mode can reach 1.53, increasing of 28.6% as compared with the COP 1.19 in the double effect mode.
전자현미경을 이용한 나노셀룰로오스 물질의 형태학적 특성 분석 연구
권오경(Ohkyung Kwon),신수정(Soo-Jeong Shin) 한국펄프·종이공학회 2016 펄프.종이기술 Vol.48 No.1
Electron microscopy is an important investigation and analytical method for the morphological characterization of various cellulosic materials, such as micro-crystalline cellulose (MCC), microfibrillated cellulose (MFC), nanofibrillated cellulose (NFC), and cellulose nanocrystals (CNC). However, more accurate morphological analysis requires high-quality micrographs acquired from the proper use of an electron microscope and associated sample preparation methods. Understanding the interaction of electron and matter as well as the importance of sample preparation methods, including drying and staining methods, enables the production of high quality images with adequate information on the nanocellulosic materials. This paper provides a brief overview of the micro and nano structural analysis of cellulose, as investigated using transmission and scanning electron microscopy.
< 구두-B-08 > 앙상블법을 이용한 자동목재수종식별 시스템의 성능향상
권오경 ( Ohkyung Kwon ),이형구 ( Hyung Gu Lee ),양상윤 ( Sang-yun Yang ),김현빈 ( Hyunbin Kim ),박세영 ( Se-yeong Park ),최인규 ( In-gyu Choi ),여환명 ( Hwanmyeong Yeo ) 한국목재공학회 2019 한국목재공학회 학술발표논문집 Vol.2019 No.1
이전 연구에서 개발된 딥러닝 기술과 제재목의 횡단면 이미지를 이용한 목재수종 자동판별 모델인 LeNet3는 목재 횡단면 이미지에 대해 높은 판별 성공률을 보였다. 하지만, 실제 현장에서 활용할 때에는 판목면의 이미지를 얻게 될 경우가 더 많다. 따라서 현장에서의 목재수종 자동판별의 성능을 향상시키기 위해서는 판목면 이미지를 대상으로 하는 모델의 개발이 필요하다. 침엽수재의 판목면 상의 무늬는 횡단면에 비해 수종 간 차이가 덜 명확하며, 잘린 각도에 따라 다양하고 복잡한 무늬가 나타난다. 그 결과 횡단면의 이미지에 대한 높은 판별 성능을 보이는 모델을 적용할 경우, 판목면 이미지에 대한 판별 성능이 낮아질 것으로 예상된다. 따라서 횡단면 이미지, 판목면 이미지 모두에 높은 판별 성능을 보이는 모델을 개발할 필요가 있다. 본 연구에서는 목재의 횡단면, 판목면의 이미지를 이용하여 목재수종을 자동으로 식별하는 앙상블 방법을 이용하여 새로운 모델을 개발하였다. 앙상블 방법은 기존의 여러 가지 모델을 이용하여 분류성능을 향상시키는 방법으로 특정한 무늬에 대한 판별 성능이 높은 여러 개의 모델을 결합하는 방법이다. LeNet 계열의 모델들과 MiniVGGNet 계열의 모델들이 조합 중에서 LeNet2, LeNet3, MiniVGGNet4를 이용한 앙상블 모델의 성능이 가장 좋게 나왔다. 이 앙상블 모델을 이용하여 한국산 5개 수종의 횡단면, 판목면 이미지를 이용하여 목재수종 자동식별을 수행한 결과 개별 모델보다 판별 성능이 향상(f1 score > 0.98)된 것을 확인할 수 있었다. 특히, 잣나무와 소나무에 대한 판별 성능이 크게 향상된 것을 확인하였다.