최소 단어 이상 선택하여야 합니다.
최대 10 단어까지만 선택 가능합니다.
다음과 같은 기능을 한번의 로그인으로 사용 할 수 있습니다.
NTIS 바로가기다음과 같은 기능을 한번의 로그인으로 사용 할 수 있습니다.
DataON 바로가기다음과 같은 기능을 한번의 로그인으로 사용 할 수 있습니다.
Edison 바로가기다음과 같은 기능을 한번의 로그인으로 사용 할 수 있습니다.
Kafe 바로가기주관연구기관 | 한국온실작물연구소 |
---|---|
연구책임자 | 서범석 |
참여연구자 | 박양호 , 윤두현 , 정두남 , 이정주 , 서수원 , 임정희 , 김희곤 , 양원모 , 이옥정 , 박철수 , 안상진 , 박경섭 |
보고서유형 | 최종보고서 |
발행국가 | 대한민국 |
언어 | 한국어 |
발행년월 | 2016-12 |
과제시작연도 | 2016 |
주관부처 | 농촌진흥청 Rural Development Administration(RDA) |
등록번호 | TRKO201700006483 |
과제고유번호 | 1395048063 |
사업명 | ICT융합 한국형 스마트팜 핵심기반기술개발 |
DB 구축일자 | 2017-09-20 |
DOI | https://doi.org/10.23000/TRKO201700006483 |
Ⅳ. 연구개발 결과
1. 토마토 엽 이미지를 이용한 엽면적 측정 알고리즘
○ 엽폭과 엽장의 곱에 0.247을 곱한 직선회귀식에 의한 토마토 개별 엽면적 추정은 높은 결정계수와 잔차 표준화에 따른 개별 엽면적은 정규분포에 더욱 가까운 값으로 추정되어 엽장이나 엽폭을 단독으로 사용하는 것보다 가장 합리적인 모델로 인정되었다.
○ 엽위별 엽형지수를 이용하여 군락내에서 이미지 확인이 가능한 엽장, 엽폭 중 하나를 이용하여 엽면적을 측정하는 데 이용할 수 있으며, 최초의 품종별 엽장, 엽폭, 엽형지수,엽면적을 픽셀로 분석하여
Ⅳ. 연구개발 결과
1. 토마토 엽 이미지를 이용한 엽면적 측정 알고리즘
○ 엽폭과 엽장의 곱에 0.247을 곱한 직선회귀식에 의한 토마토 개별 엽면적 추정은 높은 결정계수와 잔차 표준화에 따른 개별 엽면적은 정규분포에 더욱 가까운 값으로 추정되어 엽장이나 엽폭을 단독으로 사용하는 것보다 가장 합리적인 모델로 인정되었다.
○ 엽위별 엽형지수를 이용하여 군락내에서 이미지 확인이 가능한 엽장, 엽폭 중 하나를 이용하여 엽면적을 측정하는 데 이용할 수 있으며, 최초의 품종별 엽장, 엽폭, 엽형지수,엽면적을 픽셀로 분석하여 품종에 따른 엽형지수의 차이를 보정할 수 있다.
2. 토마토 이미지를 이용한 생육량 분석 알고리즘
○ 온실 내에 설치된 로봇이 촬영한 영상 및 적외선 이미지를 데이터화하여 작물의 생육량을 자동으로 측정하고, 생육을 진단하는 알고리즘
○ 주요 측정 항목
- 엽 : 엽장, 엽폭, 엽면적, 엽면적지수
- 줄기 : 마디굵기, 마디길이, 초장
- 화방 : 화방수, 화방당 꽃의수
- 과일 : 과장, 과폭, 착과수, 과일색상
3. 작물생육 이미지를 이용한 생육량 분석시스템
○ 온실 내에 설치된 로봇이 촬영한 영상 및 적외선 이미지를 데이터화하여 작물의 생육량을 자동으로 측정하고, 생육을 진단할 수 있는 프로그램 개발
○ 개발제품 : 이미지분석 S/W
- 엽 : 엽장, 엽폭, 엽면적, 엽면적지수
- 줄기 : 마디굵기, 마디길이, 초장
- 화방 : 화방수, 화방당 꽃의수
- 과일 : 과장, 과폭, 착과수, 과일색상
- 온도 : 엽온, 과일온도, 줄기온도, 온실표면온도
4. 토마토 과고, 과폭을 이용한 수확과중 예측기술
○ 토마토 과실의 이론적 볼륨인 “과폭2×과고”를 계산하여 과중과의 관계는 0.0005를 곱한 직선 회귀식을 나타냈다.
5. 열화상카메라를 이용한 작물상태 모니터링 방법
○ 열화상 카메라를 이용하여 군락상단에서 식물체 엽온 열화상 촬영
- 근권수분함량이 40%이하 시 기온과 같거나 1~2℃정도 높음
○ 수경재배 토마토의 양액농도에 따라 엽온의 차이를 활용
- EC 3.0dS/m 이상에서 기온과 같거나 주변엽보다 엽온이 높음
6. 고정형 반자동 센싱 로봇 개발
○ 작물 생육량 이미지를 표준화된 매뉴얼에 따라 자동으로 측정할 수 있는 카메라 하우징기술과 로봇 및 제어관리 시스템
○ 농가 보급 제품 : 고정형 반자동로봇, 카메라 및 주변장치
7. 고정형 자동센싱 로봇 개발
○ 작물 생육량 이미지를 자동 취득하기위해 전용화된 4축 로봇 기구 및 제어부 구성을 통해 농가에 보급이 가능 하도록 최적화한 고정형 자동센싱 로봇 개발
○ 인터넷을 이용한 로봇 원격 제어와 자동 생육 정보 조사 가능
8. 작물 측정 스마트폰 앱 프로그램 개발
○ 스마트폰 APP을 통해 엽위별 픽셀별 엽면적을 측정
○ 스마트폰 APP을 통해 엽위별 엽장, 엽폭을 측정함
○ 개별 엽위별 엽장,엽폭,엽면적 측정을 위한 스마트폰 작물촬영
○ 작물측정 스마트폰 APP 프로그램 설계
9. 이미지 생육데이터 수집용 임베디드 카메라 개발
❍ 프로토 타입을 만들어 작물 이미지를 수집
❍ 작물 영상이미지 데이터를 수집을 하기 위한 알고리즘으로 TimeLapse 기능을 구현
○ 대면적 이동형 7축 작물 생육 정보 자동 센싱 로봇 개발
○ 작물 생육 정보 자동 센싱 로봇 개발
○ 작물 생육 정보 자동 센싱 로봇 대면적 이동형과 고정형에 대해 특허 2건 출원
11. 작물생육 이미지 센싱 개방형 플랫폼 개발
○ 플랫폼 UI : C++ Builder 개발
○ 플랫폼 내 환경 : 오픈소스 OS로 Winmdows 기반 플랫폼
○ 오픈소스 라이브러리
- Web server : mongooe, PHP : v5.2.1, XML' : libXML‘, 제어 : WiringPi,
- 통신 : RS232, RS 485, TCP/IP Wifi, - Webserver / Webservice I/F
( 출처 : 요약문 6p )
< 1st sub-rsearch, KGCRI >
This study carried out to estimate growth characteristics of tomato grown in the different stress conditions such as water, light, plant density, etc. Plant growth represented to the leaf area index in the study, but very difficult in the measurement of leaf area as aff
< 1st sub-rsearch, KGCRI >
This study carried out to estimate growth characteristics of tomato grown in the different stress conditions such as water, light, plant density, etc. Plant growth represented to the leaf area index in the study, but very difficult in the measurement of leaf area as affected by the plant density and overlapped leaf, especially to measuring plant growth through image was very hard. According to the reason, has developed diagnostic algorithm to estimate growth and development through the cooperative estimation using the L-system(Lindenmayer system) and leaf shape index, leaf area index collected from plant image.
The measurement of leaf area using the regression analysis based on the measured leaf length and leaf width of tomato grown in greenhouse. The coefficient of determination of fitted the linear regression to the values of leaf length multiple by leaf width was much higher than the value of the fitted to a single independent value. When planting density is increased the index(the value of leaf length divided leaf width) increase, especially leaf length was higher than leaf width of tomato plant. Although estimated a single leaf area with increase leaf position showed a large residual between measured and fitted values, a tendency suggest that few measured leaf length was long width wide leaf width. In order to make a good regression model a research would observed the elongation of the leaf width and length, season, cultivar(cherry, medium and large tomato group) for stabilizing the coefficient of the regression.
In this study carried out to estimating fruit weight using fruit width and fruit height, the measurement of fruit weight using the regression analysis based on the measured fruit width and fruit height of tomato grown in greenhouse. The coefficient of determination of fitted the linear regression to the values multiply by 0.0005 of fruit height multiple by the square of fruit width was high.
Growth image of tomato plant grown by the 5 levels of water stress, plant growth proceed from appearance of the 4th cluster to the 8th cluster, growing conditions until 5th to 6th leaves from meristem, leaf length are very highly grow, and distance until the 1st flower from meristem is the distance stressed plant is shorter than the normal one as average of 17.23cm, and flowering cluster showed how close to the meristem.
Through the plant image analysis to have measure the leaf length and leaf width at 1st to 7th leaf position from meristem and need to have the new measuring methods of eaf length and leaf width using the leaf shape index for leaves to difficult to measure the leaf length and leaf width. The leaf shape index are showed at the range of 1.54 to 1,16 according to the increase in the leaf position from meristem.
In the experiment of artificial lights scuh as MH(metal halide lamps), HPS-1(high pressure sodium vapor lamp, 250W), HPS-2(high pressure sodium vapor lamp, 400W), FL-1(general fluorescent lights), FL-2(addition of red light on the fluorescent light) on the growth and yield responses of tomato “Rapito” variety, plant growth rate and yield responses were different as according to the growing stages. In the plot of MH plant heights were very high as over of 30cm at plant height than natural light. In the leaf CGR, leaf NAR all plots of artificial lighting without LED were higher than the control until 111days after transplanting.
Plant heights and leaf area of tomato plant in the different water stress were decrease as according to reducing water contents in the growing media as proportionally, and distance from the meristem to the first flowering cluster were shorter at 17.3cm, average in the water stressed plant than the control plants. Fresh and dry weights of leaf and stem and leaf area were reduced proportionally in extent of water stress.
To measure the leaf area depends on the planting densities, image informations acquired from the leaves in order to estimate the leaf area including the leaf width, leaf length often remain that can not be identified, especially, can calculate using leaf shape index(ratio of the leaf width divided by the leaf length) as the growing stages. Especially, leaf area calculate from multiply 0.0467 in the value of leaf length multiply leaf width.
Plant growth amount through the robots which collect plant images for analysis of plant height, stem diameter, leaf area, and leaf area index, accurate with more than 99 % when comparing the value of a direct investigation.
At the nondestructive ways to predict the individual weight of fruit weight in the young fruit stage, the fruit height was increased by 1.19 times fast growth than the increase of fruit width, but in the ripening fruit stage was not increased in the fruit height. In the ripening stage, increasement of 1mm in fruit width was equal to the increasement of 6.37g in fruit weight, but was not equal to the young fruit stage. The volume of the fruit at premature stage was calculated by multiplying 0.0005 to square of fruit width and fruit height and to the linear correlation is intense.
< 1-Cowork-research, JARES >
This study was carried out to investigate the effect of water limit through the control of water content within rhizosphere on the leaf temperature changes of hydroponically grown tomato plant as a water stress index. As water contents treated were 30, 35, 40, 45, 50±5%, the leaf temperature was lower 0.5 to 1.1℃, but pericarp temperature was higher 1.8℃ than air temperature in August. On second day of water limit, leaf temperature was highest at 3 pm and higher 2.8℃ than air temperature. After about 60 days of water limit, leaf temperature as affected by below 40% water content was similar with or higher than air temperature. Therefore these results could be used as a water stress index, however, it would be considered that more investigation and analysis on leaf temperature changes by time passes were needed.
< 2-Cowork-research, Sunchon Univ. >
6 kinds of 0.3 ∼1.4M pixels camera were selected. Actual growth data at the same time as the measurement, photographs are taken, and image growth data is analyzed using a screen ruler. Error rates of Actual Measured Values and Image-Based Measured Values : stem node length of 1 to 4 node is 2.24∼17.88%, 5 to 6 node is 44.87∼51.19%. stem node diameter 1 to 5 node is 29.73∼32.24%, 6 node is 79.58%, leaf length 1 to 4 node is 14.33∼ 25.72, 5 to 6 node is 47.14∼90.77%, leaf width is 25.54∼38.94%, 5 node is 48.56%.
The error rate of 5 to 6 node is about twice the error rate of 1 to 4. It was judged to be due to the difference in the initial state change of growth of 5∼6 nodes near the growth point. In the image growth index survey, the appropriate position of the target is the leaf attached to the node of the first flower bloom at the bottom of the growth point.
Raspberrypi developed in the form of multi-purpose platforms such as greenhouse monitoring, environmental measurement, image growth data collection, data storage and transmission, and complex environmental control combined with camera module and sensor. It is designed to be monitored by user's smart phone through streaming server and web server. It collects crop image data and environmental data together, analyzes them, and displays them in graph form. A web server was installed on the embedded device so that the streaming video can be monitored on the smart phone. It installs a framework and provides images to the user through the network in the form of a web application. Implementation of login page, video monitoring page, environmental data monitoring page, data analysis page, configuration page.
Installed a distance indicator to the tomato. Taken in side and top 45 degrees with 2D embedded camera and 3D camera(3 directions, 4 times in 10 plants at every 1 week). A computer program called screen ruler, which measures the size of a pixel on a computer screen, measures the leaf length and width, stem length, and diameter(growing point below, flower cluster is an epiphytic stem node bloomed first flower is given by 1 stem node, and towards growing point sequentially 5 or 6). Stem node length error rate : 1st node is -2.7, 2nd node is 5.74, 3rd node is 5,0, 4th node is 24.7, 5th node is 41.7, finally 6th node is 38.2%. It is desirable to utilize 1st to 3rd nodes in the vicinity of the first node of the first flower blooming process. Difference in image interpretation value occurs depending on shooting direction. The reasons are leaf overlap, stem clipping, image blur, uncertain point of measure node, and so on. The difference in the manual analysis value between the 3D camera and the 2D camera image was not large. 2D(5Mpixel), which has a high number of pixels, was easy to measure because there was little distortion when enlarging pictures.
We compared the analysis data of the 3D camera image with the engine data and the manually analyzed data with the screen data by screen ruler and the actual data. The problem of 3D image growth data is derived. The analysis results of leaf length and width were obtained from 3 out of 10 individuals. Shooting direction, shooting angle, and inter-entity interference are the causes of the decrease in recognition rate. The error of screen ruler analysis was smaller than the measured value and the error value of analysis value by automatic engine was very large. In order to improve the accuracy of engine recognition, it is necessary to consider institutional characteristics that reflect the characteristics of the growth survey item, and to improve the software.
< 3-Cowork-research, Ein Information Tech. Co., Ltd >
"Crop Growth Visualization System Image Analysis and Analysis Model Software Development" research project is to automatically capture image images through fixed robots and mobile robots, which are automatic measuring devices in the room, and the automatically captured image images are classified into "crop growth analysis system(PGAS v1.00)" Leaf, stem, flower, and fruit were separated from each other, and the separated crop plants were measured for leaf length, leaf width, leaf area, number of flowers,
In addition, the average temperature of crop plants separated by thermal imaging camera during image shooting is identified and stored in a database for analysis of crop growth.
Depth Camera was used for precise measurement of crops and it was taken at the same time with the thermal camera to reduce the error according to the surrounding environment.
In order to match the angle of view between the depth camera and the thermal camera,two images of the depth camera and the thermal camera were mapped and matched.
In order to prevent the deviation of the robot from the image capturing coordinates as the crop grows, an automatic robot coordinate increment method is applied to the system so that the image capturing is possible in accordance with the crop growth.
< 5-Cowork-research, Ein Information Tech. Co., Ltd >
Development of crop monitoring system using robot. Development of fixed and mobile robotic system. Development of Internet unmanned driving system.
< 6-Cowork-research, Protected Horticulture Research Institute >
The results of demonstration test which plant growth analysis using image scanner was no problem, but, need more test for increasement of accuracy.
( 출처 : SUMMARY 14p )
과제명(ProjectTitle) : | - |
---|---|
연구책임자(Manager) : | - |
과제기간(DetailSeriesProject) : | - |
총연구비 (DetailSeriesProject) : | - |
키워드(keyword) : | - |
과제수행기간(LeadAgency) : | - |
연구목표(Goal) : | - |
연구내용(Abstract) : | - |
기대효과(Effect) : | - |
Copyright KISTI. All Rights Reserved.
※ AI-Helper는 부적절한 답변을 할 수 있습니다.