SEGMENTATION D’IMAGES DE PLANTES CAPTURÉES PAR UN SYSTÈME D’IMAGERIE FLUORESCENTE
Results and Discussion
All TIS-III hardware and software went through a month of verification and validation prior to the deployment to the Arctic greenhouse (Figure 1.9A). The output of the excitation lights (filtered light) and grow lights was verified using a USB4000 spectrometer, CC-3 cosine corrector, one meter QP400-1-UV-VIS optical fiber and SpectraSuite software from Ocean Optics (Longueuil, Quebec Canada). Measurements were taken using a calibrated FLUKE multimeter to verify the excitation lights current source output. An oscilloscope (Tektronix TDS3054B) was used to verify the PWM grow light voltage frequency, ratios and voltage output. After calibrating the lens to focus on the biological sensor and adjusting the camera settings to capture the maximum amount of fluorescence (without saturation), tests were conducted to verify that only the emission lights were captured. The described validation process was implemented to confirm that the developed imaging system met all the initial design requirements. The process confirmed that all the components where communicating and performing nominally. In addition, to minimize technology deployment risk, maximize confidence in any new modifications and to improve researchers work efficiency, every new or improved system was validated at a replica of the ACMG greenhouse at the Canadian
Space Agency (CSA) headquarters in Longueuil, Quebec (Figure 1.9B) (Paul et al., 2008). A simulated deployment was performed for the TIS-III at the CSA greenhouse and confirmed compatibility.
Upon arrival on Devon Island, several functional tests were conducted at HMP, before connecting TIS-III to the ACMG system, to ensure that nothing was damaged during shipping and handling of the system. In addition, the image sensor and lens were recalibrated using a calibration plate designed and fabricated by NASA Kennedy Space Center (KSC).
During installation, it was realized that, unlike the CSA development greenhouse, the ACMG did not have a properly regulated 24 VDC output. The 24 VDC was connected directly to the DC renewable energy system battery bank. By examining the data collected by the greenhouse sensors of the previous year, it indicated that the 24 VDC unregulated line varies between ~23.5 V to peaks of ~33 V. The imager hardware was examined to study the possibility of modifying parts not capable of withstanding the high voltage peaks. The cFP32 2120 controller data sheet indicated that the power supply range was 11–30 VDC and since similar controllers were used from the beginning of the greenhouse deployment in 2002 (Giroux et al., 2006), it was concluded that it was safe to use the unregulated 24 VDC line on the TIS-III controller. The NI-1744 camera data sheet specified that the smart camera accepted power within the range of the industry standard IEC 1311 input power specification (24 V +20%/–15% with an additional allowance for an AC peak of +5%), at 33 V (short and infrequent peaks), it’s slightly above that specified by the manufacturer. In addition, the camera was powered for only a few minutes every few hours. With this information it was also decided that it was safe to connect the camera directly to the unregulated line. Since the excitation board was connected to the camera current source (direct drive), not only the lights were safe, but their intensity would not fluctuate in the event of power variation as the camera would adjust to apply the same current set by the user. The only element that had to be further developed due to the issue of unregulated power was the grow light board and this was accomplished by increasing the ballast resistor values that limit the current. It was mentioned previously that a resistor for each row was added to limit the current to the maximum allowed in that row. Using this information, the specifications of the LEDs that limited the current in each row were consulted to determine the new ballast resistor values. It was also considered that since the voltage peaks are in bursts of short periods and as LED specifications indicated that they would withstand a higher current in short duty cycles, the
new ballast resistor values were calculated using the higher current limits. This solution decreased the maximum grow light intensity, but the intensity was still maintained within the range of the Arabidopsis requirements. After the changes were implemented, the imager was connected to the greenhouse Ethernet hub, powered by the 24 V unregulated line and became part of the greenhouse sensor suite. Instructions and information were controlled by the greenhouse computer system. Images were stored on-site in the TIS-III controller and only JPEG files were sent via satellite with the rest of the ACMG data stream. Figure 1.10 shows TIS-III during the test phase conducted at the CSA (A) development greenhouse and during actual ACMG deployment (B).
TIS-III was deployed in the ACMG in July 2010 to help understand the operational constraints and to experiment with the integration of a fully automated imaging sensor within the ACMG system for a full year without any on-site human intervention. After deployment of the imager and departure of the field crew, TIS-III was set to capture a set of images (GFP, Black, and White) every six hours and the grow lights were operated at 100% intensity for 24 h a day. From July 10 to September 24 low resolution images were collected and transferred via the greenhouse satellite connection to Simon Fraser University. During this period, the greenhouse operated in its nominal fall operations mode where only the fall crop trays were watered (spring crop trays were inactive) (Giroux et al., 2006). Figure 1.11 shows a set of images taken during the fall run from July 11 to August 15, when the plants are in full growth and with a chosen interval of five days between images to better visualize the demonstrated growth.
Following the successful operation of the imager during fall 2010, ACMG commands were sent on September 24, via satellite link, to put the greenhouse into dormant mode, which allows the system to survive over the dark and cold season (Bamsey et al., 2009a). All systems, including TIS-III, were shutdown to conserve power, with the exception of the communication system to conserve energy. On May 1st 2011, greenhouse power was restored and TIS-III was reactivated. Still, with no human presence, TIS-III resumed collecting data and images and this information was sent south. It was then discovered that something was not operating nominally, in particular while the GFP images were fine, the regular white images were black. Following an extensive troubleshooting effort conducted over the communications link it was concluded that a remote recovery was not likely. The research team determined that a small pre-field season visit would be conducted to repair the imager while at the same time replacing the biological sample contained within. This would permit a spring operations period to be conducted with the imager during the nominal ACMG spring operations phase. The imager recovery visit permitted the scientific principle investigators of the imager, who have themselves had substantial Shuttle and ISS payload experience, the opportunity to act in a contrasting role and in this instance, implement the overall research team’s developed crew and imager repair procedures (cf. the investigators aiding in directing astronauts to conduct their developed on-orbit science/hardware operations). The repair deployment was conducted by a team of two and the conducted over a one day period (early July). The deployment team carried with them procedures for troubleshooting a suite of potential failure modes, tools and hardware for these repairs and a satellite phone to communicate with the ACMG operations team in the south. During this visit the crew traced the imager failure to a blown PWM module and in tandem to its replacement, installed a 24 VDC regulator on the line feeding the imager. Confirmation of the repair
An imaging device that captures in situ biological signals and translates these signals into measures of plant stress would be a valuable diagnostic tool for bio-regenerative life support systems. Numerous systems are available for capturing fluorescence in the laboratory but are restricted to the collection of data locally. However, biological experiments in space or in remote regions will likely be autonomous and include robotic operations where telemetric commands and data collection is a necessity because of crew unattainability and safety. The development of the fully integrated imaging system that captures fluorescence images is a significant step that stretches the technical capability of sending diagnostics in a telemetric fashion from an extra-terrestrial location. TIS-III was developed with new features including the custom designed LED grow and excitation light boards, filters, data acquisition, control system and basic environmental control, all powered with a single power source readily available at the northern greenhouse. In addition, TIS-III software was coded and integrated to the ACMG operating system, which made imager hardware and software fully compatible with the ACMG systems and contributed to the successful imager deployment as part of the sensor suite in the ACMG. Also contributing to the successful deployment was the test and simulated operations of the imager within the CSA development greenhouse. The usefulness of such an integrated test, one that included both validation of hardware as well as validation of science output is one that other such remote/extreme environment hardware deployments should consider.
TIS-III was the first imager that ran autonomously in the un-crewed greenhouse, its deployment in the ACMG and its subsequent full year of operations have demonstrated the feasibility of plant diagnostic systems that transmit and receive data by satellite link allowing near-real time monitoring and control of space biology experiments and bio-regenerative life support systems.
The Haughton Mars Project Arthur Clarke Mars Greenhouse shell was donated by SpaceRef Interactive Inc. and established at the project’s Base Camp (now Haughton Mars Project Research Station) with initial sponsorship support from the Ontario Centers of Excellence (OCE) and NASA. The greenhouse facility is currently managed and operated by the Mars Institute, in partnership with the SETI Institute and Simon Fraser University. Alain Berinstain of the Canadian Space Agency was the Principal Investigator in the Arthur Clarke Mars Greenhouse from 2002 – 2011. During this time the greenhouse was supported by the Canadian Space Agency, the University of Guelph, Simon Fraser University and the SETI Institute. We thank Martin Bergeron of the Canadian Space Agency, for his help, support and valuable manuscript suggestions. We also thank Pierre Lortie, Ralph Nolting and Maxime Pepin-Thivierge of the Canadian Space Agency machine shop for contributing to the mechanical design and for aiding in prototype construction. The authors recognize and thank Trevor Murdoch for the design of the original GIS imaging system, the shell of which formed the basis of the imager in this article. This work was supported in part by NASA grants NNX09AO78G and NNX09AL96G to Robert Ferl and Anna-Lisa Paul.
MULTISPECTRAL PLANT HEALTH IMAGING SYSTEM FOR SPACE BIOLOGY AND HYPOBARIC PLANT GROWTH STUDIES
Humans face a variety of challenges as they continue to explore beyond the frontiers of planet Earth. Exploration missions are often severely constrained by launch mass and resupply considerations. The use of plants as part of life support systems continues to be explored as an approach for more sustained human presence in space. In particular, bioregenerative life support systems have been considered since the early 20th century (Wheeler, 2010). The Canadian Space Agency, University of Florida and University of Guelph have been involved in assessing the possibility of supporting human presence on the Moon and Mars by deploying greenhouses as plant production system test-beds (Bamsey et al., 2009a; Bamsey et al., 2009b). The main concept is the use of plants to regenerate the three cornerstones of human consumable requirements; air, water and food (Tamponnet et Savage, 1994). Spaceflight and other extraterrestrial environments provide unique challenges for plant life. There originates the importance of understanding the metabolic issues that can influence plant growth and development in space. Plant monitoring systems with the capacity to observe the condition of the crop in real-time within these systems would permit operators to take immediate action to ensure optimum system yield and reliability. In addition to the utilization of chlorophyll fluorescence, specific stress response genes can be tagged with reporter genes encoding a variety of fluorescent proteins, allowing gene activities, and by extension plant health, to be monitored through the fluorescence of these gene products (Plautz et al., 1996). The Transgenic Arabidopsis Gene Expression System (TAGES) is a biosensor that uses Arabidopsis thaliana fluorescence information from both naturally occurring chlorophyll red/near infrared fluorescence, as well as green fluorescence originating from the gene products of green fluorescent protein (GFP) reporter genes (Manak et al., 2002; Paul et al., 2003). Several commercial systems are available for imaging and capturing plant fluorescence, but most analytical procedures involve laboratory examination and human input. However, advanced biological experiments on orbit, the Moon, and Mars are likely to be autonomous, precluding any direct human control over the monitoring/imaging systems. Furthermore, if a mission does include a physical human presence, there are still system trade-off considerations between internal greenhouse/growth chamber operating pressure, up-mass and crew time requirements that may still dictate completely robotic and/or autonomous bioregenerative life support systems (Paul et Ferl, 2006).
A Multispectral Plant Health Imaging System (M-PHIS) would provide a considerable step forward in our capacity to monitor advanced life support crops in an autonomous manner (Baker et Rosenqvist, 2004; Ehlert et Hincha, 2008; Galston, 1992; Lichtenthaler et Babani, 2000; Manak et al., 2002). This article describes the design and development of a prototype multispectral fluorescent imaging system deployed in a hypobaric plant growth chamber at University of Guelph. The imager was designed primarily for multiband imaging of chlorophyll and protein fluorescence with the design being driven by portability and autonomous functionality considerations. The design was also novel in that it employed a commercially available liquid crystal tunable filter (LCTF) and a custom developed LED board with an independently variable grow light LED array. This prototype imager provided real-time data while it was operated within a low pressure chamber through the use of a controller, a smart camera, and a custom designed and variable outputs grow and excitation light emitting diode (LED) light array. The deployment in a low-pressure chamber represents one of a number of possible space analogue and on orbit deployment scenarios.
The results of this work will direct future efforts in this area of research and drive further design improvements.
Table des matières
REVUE DE LA LITTÉRATURE
CHAPITRE 1 DEPLOYMENT OF A FULLY-AUTOMATED GREEN FLUORESCENT PROTEIN IMAGING SYSTEM IN A HIGH ARCTIC AUTONOMOUS GREENHOUSE
1.3 Materials and Methods
1.3.1 Fundamental Excitation System Requirements
1.3.2 Fundamental Emission Image Capture Requirements
1.3.3 Arthur Clarke Mars Greenhouse
1.4 GFP Imager
1.4.3 Grow Lights
1.4.4 Fluorescent Excitation Lights
1.5 Results and Discussion
CHAPITRE 2 MULTISPECTRAL PLANT HEALTH IMAGING SYSTEM FOR SPACE BIOLOGY AND HYPOBARIC PLANT GROWTH STUDIES
2.3 Materials and Methods
2.3.1 Excitation, Emission and Imaging Capture Requirements
2.3.2 Hypobaric Plant Growth Chambers
2.4 Multispectral Plant Health Imaging System
2.4.2 Grow Lights
2.4.3 Fluorescent Excitation Lights
2.4.4 M-PHIS Software
2.4.5 Multispectral Plant Health Imaging System to Low-Pressure Chamber Interface
2.5 Results and Discussion
2.5.1 Short Duration Low Pressure Run
2.5.2 Long Duration Low Pressure Run
CHAPITRE 3 SEGMENTATION D’IMAGES DE PLANTES CAPTURÉES PAR UN SYSTÈME D’IMAGERIE FLUORESCENTE
3.5 Bayes quadratique
3.6 Algorithme du KPPV
3.7 Machines à vecteurs de support
3.8 Comparaison des trois classifieurs
LISTE DE RÉFÉRENCES BIBLIOGRAPHIQUES