Make Your Factory Smart with Smart APS

조기환 대한산업공학회 2019 대한산업공학회 추계학술대회논문집 Vol.2019 No.11

Iterative Trimming Algorithm을 이용하여 자동추출된 KOMPSAT-3A 훈련자료 신뢰성 평가

조기환,정종철 국토연구원 2019 국토연구 Vol.103 No.-

최근까지 영상분류에서 핵심적인 과정이라 할 수 있는 훈련자료 선정이 주로 수작업에 의존하여 특징적인 지점을 선택하는 방식으로 진행되고 있다. 본 논문에서는 과거에 구축된 공간정보를 바탕으로 Iterative Trimming Algorithm을 이용한 자동화된 훈련자료 추출 기법을 제안하고 활용 가능성에 대해 검정하였다. 2015년에 발표된 세종시 세분류 토지피복지도 정보를 토대로 2018년 KOMPSAT-3A 위성영상 분류를위한 훈련자료를 추출하였으며 이를 이용하여 토지피복 분류를 실시하였다. 이를 위해 토지 유형별 확률분포를 커널밀도추정을 통해 추정하고 확률이 낮은 자료를 이상치로 간주하여 제거하는 방식으로 훈련자료를 선별하였다. 부트스트랩을 통해 산출된 토지유형별 이상치 제거 비율은 토지 유형에 따라 다르게 나타났다. 이상치 제거 비율은 시가화지역의 경우0.08, 농업/초지는 0.16, 산림은 0.04, 나대지는 0.16, 수역은 0.04일 때 가장 높을 분류 정확도를 보였다. 이상치 제거 후 선택된 훈련자료를 이용하여 토지피복 분류를 실시하고 정확도를 검정하였다. 최대우도법 분류에 대한 정확도 검정 결과 전체정확도는 약80%로 나타났고 토지유형별 분류 정확도에는 편차가 있었다. 본 논문을 통해 인위적인 자료수집과정없이 과거에 제작된 GIS 자료를 활용하여 훈련자료를 수집하고 이를 이용하여 고해상도 영상을 분류할수 있다는 것을 확인하였다. Image classification is one of the key issues of remote sensing technology and selecting training data is an essential process in supervised image classification. Dramatically increasing imagery data require more effective and automated classification techniques. The traditional process of selecting training data requires intensive manpower and, as a result, it has been costly and time-consuming. This study proposed an automatic training data extraction technique using outdated geographic information system (GIS) data and its applicability was tested. We used a high-resolution KOMPSAT-3A satellite image taken on July 7, 2018, and the land cover map in 2015 for the test of automated training data extraction based on the iterative trimming algorithm. First, the training data were extracted based on the polygon of the land cover map. Then, the probability distributions of each land cover class were estimated using kernel density estimation. The outliers were removed in the order of low probability. The bootstrap technique was used to determine the ratio of removing outliers. The ratios were different among the land cover classes. The removing ratio was 0.08 for the urbanized area, 0.16 for agriculture/land, 0.04 for forests, 0.16 for bare soil and 0.04 for water. With the refined training data, image classification was conducted. This approach allows automatic extraction of training data based on GIS data without manual digitizing. It is expected to contribute to an automatic and timely update of the urban land cover map with high-resolution imagery.

자율주행 차량의 사각지대를 감소시키기 위한 다중 라이다 배치 최적화 방법

조기환,박태형 제어로봇시스템학회 2023 제어로봇시스템학회 국내학술대회 논문집 Vol.2023 No.6

심층강화학습 기반 후방차량 감속을 고려한 차선변경 판단 기법

조기환,박태형 제어·로봇·시스템학회 2022 제어·로봇·시스템학회 논문지 Vol.28 No.6

This paper proposes a lane change judgment method based on deep reinforcement learning considering rear vehicle deceleration. In existing autonomous driving, the lane change problem focuses on maintaining speed or avoiding collisions. However, an accident may occur if the deceleration of the vehicle behind is not considered when changing lanes. In this paper, the network was designed to solve the lane change problem by applying the DQN (Deep Q-Network) and the Deep Sets network. Moreover, the compensation was designed based on the maintained speed and deceleration of the rear vehicle. A four-lane vehicle-only road environment was implemented as a simulation and compared with the existing method. The proposed method maintained speed and lowered the deceleration rate of the rear vehicle compared with the existing method. .

테니스대회를 마치고

조기환 釜山醫師會 1979 釜山醫師會誌 Vol.15 No.10

Yoghurt와 Koumiss를 급여한 Rat 血液中의 r-globulin과 Cholesterol의 변화에 관한 연구

曺驥煥,金東伸 COLLEGE OF AGRICULTURE KYUNGPOOK NATIONAL UNIVERSI 1985 慶北大農學誌 Vol.3 No.-

This study was carried out to find out changes of r-globulin and cholesterol of rat blood fed on yoghurt and Koumiss. Yoghrt and koumiss were manufactured with fortifed milk and Lactobacillus bulgaricus, streptococcus thermophilus and Saccharomyces fragilis were used. The twenty rats were devised into 4 groups with 5 replications by completely randomized design. The experimental groups are the control, milk, yoghrt and koumiss feeding groups. The results are summerized as follows ; The changes of pH after 8hrs incubation with Lactobacillus bulgricus, Streptococcus thermophilus and the mixed strains were 3, 7, 4, 6 and 3, 5 at 42℃, respectively. Average alcohol percentage of Koumiss was 1, 2(%). The average viscosity of yoghurt and Koumiss with milk showed 1500cp and 390cp. respectively at 11% of milk total solid. r-globulin contents in blood of rat fed on yoghurt and Koumiss were higher than those of control and milk. Cholesterol of rat blood in yoghurt and Koumiss group were lower than those of control and milk group.

모악산의 도깨비 왕관을 벗겨주자!

조기환 열린전북 2007 열린전북 Vol.2007 No.10

LASER(CO₂)光에 依한 各種 子宮頸部質患의 治療에 關한 臨床的硏究

趙基煥,丘秉參,洪性鳳 고려대학교 의과대학 1989 고려대 의대 잡지 Vol.26 No.1

Patients with various cervical lesion in 568 cases were treated using CO₂ Laser. The clinical results of this treatment are as follows. 1. Indications for treatment with CO₂ Laser showed the following distributions: 398 cases of cervicitis and/or ulcerative erosion, 23 cases of polyp, 4 cases of myoma, 116 cases of CIN, 8 cases of condylome, 3 cases of endometriosis, 5 cases of stenosis, and 11 cases of inclusion cyst. 2. The age distribution of the patients treated with CO₂ Laser ranged front 20 to 67 years. The peak incidence of the patients with CIN was in the fourth decade (38.8%), and ranged from the third to sixth decade. The parity distribution was as follows: nullipara, 9.5%; para one, 19.8%; and the highest incidence, 42.2% was recorded for women with a parity of three or more 3. The duration of healing after CO₂ Laser treatment varied slightly according to the disease and extent of tissue evaporation. The mean healing time was 3 to 4 weeks for cases of superficially evaporated wounds, however, the healing process was delayed up to 4-6 weeks in deeply evaporized wounds. The healing of very shallow wounds, like simple polyp excision, was rapid (2-3 weeks). 4. The vaporized wound surface, particularly in CIN, was covered with thin, gray-whitish necrotic tissue completely within post-treatment 2nd day, and the vaginal discharge was increased significantly. The amount of vaginal discharge was progressively reduced after 90 hours. The necrotic eschar began to disappear from the vaporized surface partially, and the discharge was reduced remarkably with the growing out of granulation tissue after 7 to 10 posttreatment day. Squamous metaplasia showed up after a full 2 week posttreatment period and the vaporized area became smooth and was completely covered by regenerated epithelium. 5. Amoag the total 116 CIN cases, 67 cases (58%) were class Ⅰ, 37 cases (32%) were class Ⅱ, and 12 cases (10%) were Class Ⅲ. The lesions were vaporized more deeply (33%), if they were located in the cervical canal or if the grade was severe. 6. During the follow up period of 36 months in CIN cases, treatment failure confirmed by cytologic examination in class Ⅰ was 2 cases, in Ⅱ and Ⅲ were 3 respectively. Retreatment with Laser was performed in 4 (1 in class Ⅰ, 2 in Ⅱ and 1 in Ⅲ) of the 8 initially failed casea, and others were treated surgically. 7. The success rate for CO₂ Laser therapy in various cervical lesions was 94.8% (recurred in 5.2%). The treatment success rate for CIN cases was 91% (Grade Ⅰ; 98.5%, Ⅱ; 91.5%, and Ⅲ; 83.3%). 8. Time required for the operative procedure varied depending on the physician, his experience, and the extent or location of the lesions. Except for Laser conization or hemorrhagic lesions, 5 to 10 minutes were enough to complete the therapy in the majority of cases (70%), 15 to 20 minutes were required in 20% of cases, and rarely required more than 25 minutes in unusual cases of conization or severe hemorrhagic lesions. 9. No specific anesthetics and analgesics were necessary in the majority of cases. Intravenous analgesics were used in 15 cases, and general anesthesia was needed (3 cases) in some cases of Laser conization. The percentage that needed any type of analgesics including local infiltration or nerve block with local analgesics, or intravenous pethidine was 8%. 10. There were no major complications. 8 cases (1.4%) of mild to moderate bleeding and 3 cases (0.5%) of infection were reported, and a mild degree of local or lower abdominal discomfort was the chief complaint in 3.2% of patients. However, no blood transfusion or hospitalization were necessary. 11. Laser therapy for various cervical lesions has many advantages compared with other surgical modalities (electrocautery, cryosurgery, or cold conization etc.) that is, simplicity in performance, accurate application, both with regard to site and depth, and rapid healing process. These results indicate that the CO₂ Laser therapy, when performed by a well trained and experienced physician, is an effective modality in the management of various cervical lesions.

토지피복 공간정보를 활용한 자동 훈련지역 선택 기법

조기환,정종철 한국국토정보공사 2018 지적과 국토정보 Vol.48 No.2

Despite the rapid expansion of satellite images supply, the application of imagery is often restricted due to unautomated image processing. This paper presents the automated process for the selection of training areas which are essential to conducting supervised image classification. The training areas were selected based on the prior and cover information. After the selection, the training data were used to classify land cover in an urban area with the latest image and the classification accuracy was valuated. The automatic selection of training area was processed with following steps, 1) to redraw inner areas of prior land cover polygon with negative buffer (-15m) 2) to select the polygons with proper size of area (2,000 ~200,000㎡) 3) to calculate the mean and standard deviation of reflectance and NDVI of the polygons 4) to select the polygons having characteristic mean value of each land cover type with minimum standard deviation. The supervised image classification was conducted using the automatically selected training data with Sentinel-2 images in 2017. The accuracy of land cover classification was 86.9% ( = 0.81). The result shows that the process of automatic selection is effective in image processing and able to contribute to solving the bottleneck in the application of imagery. 급속한 위성영상 공급확대에도 불구하고 자동화되지 못한 영상처리과정으로 인해 영상활용이 제약받는 경우가 많다. 본 연구에서는 감독영상분류를 위한 훈련지역 선택과정을 자동화함으로써 영상처리과정의 비용과 시간을 절감하는 방안을 제시하였다. 이를 위해 기존의 토지피복 정보를 활용하여 훈련관심영역을 추출하는 방법을 최신영상에 적용하여 토지피복분류를 실행한 후 분류정확도를 평가하였다. 원주시 도심지역을 대상지로 하여 토지유형을 시가지역과 농지/초지, 숲, 나대지 및 수계로 나누고 유형별 훈련관심영역을 환경부 중분류 토지피복지도를 활용하여 선택하였다. 관심영역 선택을 위해 먼저 토지피복지도 폴리곤 경계를 기준으로 negative buffer (-15m)를 적용하여 새로 폴리곤을 만들었고 너무 작은 폴리곤(<2,000㎥)과 큰 폴리곤(>200,000㎥)을 제외하였다. 선택된 폴리곤들의 밴드별 반사율 표준편차와 평균값 및 NDVI의 평균값을 계산하였다. 이 정보를 이용하여 먼저 표준편차가 적은 폴리곤 (폴리곤 내 반사율 값의 편차가 크지 않은 폴리곤)을 선택한 후 이들 중 반사율 평균값이 각 유형의 특징적인 분광특성을 반영할 수 있는 폴리곤을 관심영역으로 선택하였다. 2017년 Sentinel-2영상을 활용하여 토지피복유형을 분류한 결과 86.9%의 분류정확도(=0.81)가 도출되었다. 본 연구에서 시도된 자동 관심영역 선택방법 적용한 결과 수동 디지타이징 과정을 생략하고도 높은 분류정확도를 도출 할 수 있었으며 이와 같은 방법을 통해 영상처리에 필요한 시간과 비용을 절약하여 급속히 증가하고 있는 영상을 효율적으로 활용할 수 있게 될 것으로 기대된다.

오프라인 템플릿을 이용한 RRT<SUP>*</SUP> 기반 동작 계획

조기환,박태형 제어로봇시스템학회 2020 제어로봇시스템학회 국내학술대회 논문집 Vol.2020 No.7