http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
부동 소수점 연산을 이용한 실시간 영상 편위교정 FPGA 하드웨어 구조 설계
한동일(Dongil Han),최재훈(Jeahoon Choi),신호철(Ho Chul Shin) 대한전자공학회 2014 전자공학회논문지 Vol.51 No.2
본 논문은 두 대의 카메라로 찍은 영상을 이용하여 사물의 3D 정보를 계산하는 스테레오 매칭(Stereo Matching) 기법의 전처리 과정에 관한 연구이다. 본 논문에서는 카메라 내부의 왜곡 및 두 카메라간의 정렬 문제로 인해 생긴 영상의 수직시차(vertical parallax)를 제거하기 위한 실시간 편위교정(Rectification) 하드웨어 설계 구조를 제안한다. 이를 위한 사전 단계로 J.Y Bouguet이 설계한 Matlab 툴박스를 이용해 영상의 보정 매개변수(calibration parameter)를 구한 후 Heikkila 와 Silven의 알고리즘을 기반으로 하여 편위교정 하드웨어를 설계하였다. 이때 결과 이미지의 정밀도를 높이기 위하여 Xilinx의 Core 생성기를 이용해 부동소수점 연산기를 생성하여 사용하였으며, 이를 통하여 룩업 테이블(Look-Up Table) 등을 사용하여 설계된 타 편위교정 하드웨어에 비해 높은 정밀도를 가지면서도 실시간으로 작동하는 편위교정 하드웨어를 설계할 수 있음을 확인하였다. This paper suggests a novel hardware architecture of a real-time rectification which is to remove vertical parallax of an image occurred in the pre-processing stage of stereo matching. As an off-line step, Matlab Toolbox which was designed by J.Y Bouguet, was used to calculate calibration parameter of the image. Then, based on the Heikkila and Silven"s algorithm, rectification hardware was designed. At this point, to enhance the precision of the rectified image, floating-point unit was generated by using Xilinx Core Generator. And, we confirmed that proposed hardware design had higher precision compared to other designs while having the ability to do rectification in real-time.
연구논문 : 친환경 태도와 친환경의류제품 구매에 따른 여성 소비자의 유형화 및 특성
한동일 ( Dongil Han ),김준호 ( Junho Kim ),나영주 ( Youngjoo Na ) 한국의류산업학회 2013 한국의류산업학회지 Vol.15 No.5
This study divides female consumers according to eco-friendly attitudes and the purchase frequency of ecoclothing products; in addition, it analyzes the characteristics of each group in terms of LOHAS lifestyles, the attitudes for eco-clothing products, and satisfaction. Eco-clothing attitudes of female consumers were lower than eco-friendly attitudes. A total of 360 female consumers were divided into 4 group according to purchase frequency and the eco attitude, Type 1: survival (33.0%), Type 2: wellbeing (25.6%), Type 3: curious (15.0%), Type 4: LOHAS (26.4%); in addition, age, income, marit alstatus, occupation, LOHAS lifestyle and shopping places were differentiated by type. Type 1 (low eco attitude and low eco-clothing purchase) were composed of the youngest, college students, low income, low level of LOHAS lifestyles and low level of eco-clothing attitude. Type 2 (high eco attitude and low eco-clothing purchase) were interested in healthy eating & exercise among LOHAS lifestyles; in addition, Type 1 & 2 showed alow level of eco-clothing satisfaction. Type 3, low eco attitude and high eco-clothing purchase, were characterized with high expenditures on clothing relative to income as well as lower levels of family activities and leisure life than LOHAS Type 4. Type 4 (high eco attitude and high eco-clothing purchase) were the oldest group and mostly composed of married workers (the highest income) with the highest LOHAS lifestyles and the highest level of eco-clothing satisfaction.
이진 까마귀 탐색 알고리즘을 활용한 고추 병해 진단 연구
한동일(Dongil Han),김지선(Jisun Kim),최다빈(Dabin Choi) 대한전자공학회 2021 대한전자공학회 학술대회 Vol.2021 No.6
We study an algorithm for diagnosing diseases occurring in peppers using deep learning technology based on NAS that automatically discovers artificial neural network architectures. It compares and analyzes the diagnosis result from the algorithm other than the NAS and the experiment result by applying the disease data to the existing NAS and proposes a new algorithm by further developing the BCSA algorithm. It has been proved that the performance of the classifier can be improved through genetic algorithms and the optimal architecture for diagnosing diseases can be found through genetic algorithms without humans directly adjusting the parameters. It allows you to explore the architecture in space. In this study, an experiment to diagnose pests and pests of crops was conducted based on images, and an accuracy of 99% was achieved.
한동일(Dongil Han),최종호(Jongho Choi),유성준(Seong Joon Yoo),오세창(Sechang Oh),조재일(Jae-Il Cho) 大韓電子工學會 2011 電子工學會論文誌-SP (Signal processing) Vol.48 No.4
본 논문에서는 기존의 방법에 비해서 사용되는 메모리의 증가가 없이, 혹은 메모리의 증가를 최소화하는 영상 메모리의 회전 변환 기법을 개발하여 얼굴 회전 변화에 강인한 고성능 실시간 얼굴 검출 엔진 구조를 제안하였으며 FPGA 구현을 통하여 제안 구조의 타당성을 검증하였다. 고성능 얼굴 검출을 위해 기존에 사용하던 조명 변화에 강인한 MCT(Modified Census Transform) 변환 기법과 최적화된 학습데이터 생성을 위한 Adaboost 학습 기법 이외에 얼굴 회전 변환에 강인함을 위한 영상 회전 기법을 이용하였다. 제안한 하드웨어 구조는 색좌표 변환부, 잡음 제거부, 메모리 인터페이스부, 영상 회전부, 크기 조정부, MCT 생성부, 얼굴 후보 검출부/ 신뢰도 비교부, 좌표 재조정부, 데이터 검증부, 검출 결과 표시부/ 컬러 기반 검출 결과 표시부로 구성되어있다. 구현 및 검증을 위해 Virtex5 LX330 FPGA 보드와 QVGA급 CMOS 카메라, LCD Display를 이용하였으며, 다양한 실생활 환경 및 얼굴 검출 표준 데이터베이스에 대해서 뛰어난 성능을 나타냄을 검증하였다. 결과적으로 실생활 환경에서 초당 60프레임 이상의 속도로 실시간 처리가 가능하며, 조명 변화 및 얼굴 회전 변화에 강인하고, 동시에 32개의 다양한 크기의 얼굴 검출이 가능한 고성능 실시간 얼굴 검출 엔진을 개발하였다. In this paper, we propose the structure of a high-performance face-detection engine that responds well to facial rotating changes using rotation transformation which minimize the required memory usage compared to the previous face-detection engine. The validity of the proposed structure has been verified through the implementation of FPGA. For high performance face detection, the MCT (Modified Census Transform) method, which is robust against lighting change, was used. The Adaboost learning algorithm was used for creating optimized learning data. And the rotation transformation method was added to maintain effectiveness against face rotating changes. The proposed hardware structure was composed of Color Space Converter, Noise Filter, Memory Controller Interface, Image Rotator, Image Scaler, MCT(Modified Census Transform), Candidate Detector / Confidence Mapper, Position Resizer, Data Grouper, Overlay Processor / Color Overlay Processor. The face detection engine was tested using a Virtex5 LX330 FPGA board, a QVGA grade CMOS camera, and an LCD Display. It was verified that the engine demonstrated excellent performance in diverse real life environments and in a face detection standard database. As a result, a high performance real time face detection engine that can conduct real time processing at speeds of at least 60 frames per second, which is effective against lighting changes and face rotating changes and can detect 32 faces in diverse sizes simultaneously, was developed.