http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
金漢烈,金仁澤 明知大學校 産業技術硏究所 2003 産業技術硏究所論文集 Vol.22 No.-
Abstract - This paper presents a method for detecting skin tumors on chicken carcasses using multi-spectrum image. It utilizes both fluorescence and reflectance image information in multi-spectrum images. A detection system that is built on this concept can increase detection rate and reduce processing time, because the procedure for detection can be simplified. Chicken carcasses are examined first using band radio FCM information of fluorescence image and it results in candidate regions for skin tumor. Next classifier selects the real tumor spots using PCA components information of reflectance image from the candidate regions. For the real world application, real-time processing is a key issue in implementation and the proposed method can accommodate the requirement by using a limited number of features to maintain the low computational complexity. Nevertheless, it shows favorable results and, in addition, uncovers meaningful spectral bands for detecting tumors using multi-spectrum image. The method and findings can be employed in implementing customized chicken tumor detection systems.
이종우,김한열,김신락,한예식,박진형 대한성형외과학회 2015 Archives of Plastic Surgery Vol.42 No.1
In perforator flap reconstruction, vascular mapping using preoperative computed tomography(CT) angiography is widely used to confirm the existence and location of an appropriateperforator. This study proposes a rapid, accurate, and convenient method for marking theperforator location on the skin surface. For 12 patients who underwent perforator flapreconstruction between November 2011 and November 2013, metal clips were fixed on theskin surface at the anticipated perforator locations, which were decided using a handheldDoppler. CT angiography was used to compare the location between the metal clip and theactual perforator. The metal clip was moved and repositioned, if needed, on the basis of the CTimages. The locations of the appropriate perforator and the metal clip, which were observedduring the surgery, were then compared. In CT angiography, the mean distance between themetal clip and the perforator was 3±3.9 mm, and the mean distance that was measuredduring surgery was 0.8±0.8 mm. In conclusion, we report a simple, rapid, and precisetechnique to indicate the accurate location of the appropriate perforator on the skin surface.
다중 스펙트럼 CNN 기반 고해상도 컬러 영상 보간 기법
고성희,양수미,김한열,곽성범 한국컴퓨터교육학회 2024 컴퓨터교육학회 논문지 Vol.27 No.3
본 연구에서는 다중 스펙트럼 컨볼루션 신경망(CNN)을 활용하여 고해상도 컬러 이미지 보간을 수행하는 혁신적인 기법을 제안한다. 이 방법은 저해상도 이미지를 고해상도로 변환하는 과정에서 발생하는 색상 왜곡과 디테일의 손실을 최소화하고자 한다. 우리의 CNN 모델은 다양한 스펙트럼 데이터를 입력으로 사용하여 이미지 내의 복잡한 텍스처와 색상의 상관관계를 더 정확하게 파악한다. 이를 통해 더욱 선명하고 자연스러운 색상을 가진 고해상도 이미지를 생성한다. 개선된 성능을 달성하기 위해 초해상도 네트워크는 일반적으로 많은 수의 계층과 매개변수를 포함하며, 이는 모바일 기기에서의 적용을 제한한다. 이 문제를 해결하기 위해, 우리는 모바일 임베디드 시스템을 위한 경량 이미지 초고해상도를 위한 중첩 백 프로젝션 피드백 네트워크를 제안한다. 첫째, 백 프로젝션 피드백 블록과 연결 피드백을 사용하여 네트워크의 다양한 레벨에서의 상세 특징을 효율적으로 학습한다. 둘째, 경량 네트워크에 적합한 중첩 백 프로젝션을 제안하여 복원 오류를 최소화한다. 마지막으로, 융합 주목 모듈을 제안하여 정보가 풍부한 특징에 더 많은 주의를 기울인다. 테스트 결과, 제안된 기법은 기존의 단일 스펙트럼 기반 방법들과 비교하여 향상된 시각적 품질과 색상 정확도를 보여주었다. In this study, we propose an innovative technique to perform high-resolution color image interpolation using a multispectral convolutional neural network (CNN). This method seeks to minimize color distortion and loss of detail that occur in the process of converting low-resolution images to high resolution. Our CNN model uses diverse spectral data as input to more accurately identify complex texture and color correlations within images. This creates high-resolution images with clearer, more natural colors. To achieve improved performance, super-resolution networks typically include a large number of layers and parameters, which limits their application in mobile devices. To solve this problem, we propose a nested back-projection feedback network for lightweight image super-resolution for mobile embedded systems. First, it uses back-projection feedback blocks and connection feedback to efficiently learn detailed features at various levels of the network. Second, we minimize restoration errors by proposing nested back projection suitable for lightweight networks. Finally, we propose a fusion attention module to pay more attention to information-rich features. Experimental results showed that the proposed technique showed improved visual quality and color accuracy compared to existing single spectrum-based methods.
A Rabbit Model of Fat Graft Recipient Site Preconditioning Using External Negative Pressure
이정우,한이석,김신락,김한열,김현,박진형 대한성형외과학회 2015 Archives of Plastic Surgery Vol.42 No.2
Background Fat is widely used in soft tissue augmentation. Nevertheless, it has an unpredictably high resorption rate. Clinically, external expansion with negative pressure is used to increase fat graft survival. In this study, fat graft recipient sites were preconditioned by external application of negative pressure in order to test for improvements in vascularity and fat graft survival. Methods Negative pressure was applied randomly to either the left or right dorsal ear of 20 New Zealand male white rabbits at a pressure of –125 mm Hg. The negative pressure was removed one week after the skin perfusion was measured. The skin flap at each ear was elevated, and 1 g of fat was grafted above the dorsal perichondrium. After one week, the fat weight, microvessel density, mature vessel density of the skin and fat, and amount of glycerol released were measured. Three months after the grafting, the same measurements were performed, with the exception of glycerol release. Results The fat survival rate of the experimental group (75.4%±3.9%) was higher than that of the control group (53.1%±4.3%) (P<0.001). Skin perfusion was higher in the experimental group. The glycerol release in the experimental group was significantly higher than in the control. The microvessel density of the skin and fat was significantly higher in the experimental group. Three months after the grafting, the skin and fat mature vessel density was significantly higher in the experimental groups. Conclusions Negative pressure prior to fat grafting increased the vascularity of the recipient site, and, accordingly, enhanced fat graft survival.