72 documents found in 218ms
# 1
Vakalopoulou, Maria • Karantzalos, Konstantinos
Abstract: During the last decade a significant amount of research and development has been focused on imaging spectrometry and a number of lightweight, with low power consumption, frame or pushbroom sensors have been designed to operate onboard Unmanned Aerial Vehicles (UAVs). For frame-type sensors the first essential step in the processing chain is to co-register all the acquired spectral bands. This is not a trivial task when the goal is efficiency and full automation. In this paper, we propose an algorithm for the automated co-registration of frame hyperspectral data. The operated wavelength is divided in an appropriate number of spectral groups. Then the co-registration of all the spectral bands of each group is performed. In particular, for each group the spectral band with the broader spectral variation is used as the reference for all the other ones. In a similar manner, but taking into account both the spectral variation and the proximity of each spectral band, the appropriate bands are selected for the co-registration of each group. We narrow effectively the search space of solutions by selecting with an unsupervised manner the spectral bands which can lead quickly to the solution. The developed algorithm has been evaluated both qualitatively and quantitative over agricultural fields which in contrast to urban areas their smooth terrain structure and texture impede the performance of local descriptors. Experimental results and the overall validation indicate the effectiveness of the developed algorithm.
# 2
Tilly, Nora • Hoffmeister, Dirk • Aasen, Helge • Brands, Jonas • Bareth, Georg
Abstract: Research in the field of precision agriculture is becoming increasingly important due to the growing world population whilst area for cultivation remains constant or declines. In this context, methods of monitoring in?season plant development with high resolution and accuracy are necessary. Studies show that terrestrial laser scanning (TLS) can be applied to capture small objects like crops. In this contribution, the results of multi-temporal field campaigns with the terrestrial laser scanner Riegl LMS-Z420i are shown. Four surveys were carried out in the growing period 2012 on a field experiment where various barley varieties were cultivated in small-scale plots. In order to measure the plant height above ground, the TLS-derived point clouds are interpolated to generate Crop Surface Models with a very high resolution of 1 cm. For all campaigns, a common reference surface, representing the Digital Elevation Model was used to monitor plant height in the investigated period. Manual plant height measurements were carried out to verify the results. The very high coefficients of determination (R² = 0.89) between both measurement methods show the applicability of the approach presented. Furthermore, destructive biomass sampling was performed to investigate the relation to plant height. Biomass is an important parameter for evaluating the actual crop status, but non-destructive methods of directly measuring crop biomass do not exist. Hence, other parameters like reflectance are considered. The focus of this study is on non-destructive measurements of plant height. The high coefficients of determination between plant height and fresh as well as dry biomass (R² = 0.80, R² = 0.77) support the usability of plant height as a predictor. The study presented here demonstrates the applicability of TLS in monitoring plant height development with a very high spatial resolution.
# 3
Caris, Michael • Stanko, Stephan
Abstract: The airborne monitoring of civilian and military scenes (using unmanned aircraft) is becoming increasingly important. Several types of airborne sensors – in the optical, infrared or millimeter wave spectrum - are available for the different platforms. Beside the all-weather suitability of the sensors, the recent deployment scenarios, often in deserts or arid environments, also demand for the ability to look through dust clouds and sandstorms. The only sensor, which is capable to cope with such environmental restrictions and is able to deliver high-resolution images, is the synthetic aperture radar (SAR). In this paper we focus on miniaturized SAR systems which were developed and optimized for utilization in a UAV (unmanned aerial vehicle) with a low loading capacity. This not only requires a compact and light radar sensor but the processing also has to cope with the unstable flight conditions of a small aircraft. Therefore, a high-precision inertial measurement unit (IMU) and motion compensating SAR-algorithms are needed. Thanks to the utilization of a high transmit frequency of either 35 GHz or 94 GHz, the sensors are suitable for the detection of small-scale objects and a very high resolution of 15 cm x 15 cm can be achieved when used in combination with modern FMCW (frequency modulated continuous wave) generation with a high bandwidth (up to 1 GHz) and small antennas.
# 4
Yuan, Fei • Wang, Cuizhen • Gardner, Christopher • Nagel, Philipp
Abstract: Plant Phenology refers to periodic events in the life cycle of plants as influenced by the environment, especially climatic factors. Compared to field-based plant phenological observations, remote sensing provides an alternative means for monitoring Land Surface Phenology (LSP) over large areas. Upto- date and accurate LSP information detected by remote sensors provides important inputs for assessing trends in vegetation development and ecosystem dynamics in response to global environmental changes. However, direct comparison of LSP with in situ observations is usually impractical due to the lack of pure pixel-size, species-level observation data. In this research, the methods and problems for extracting and assessing LSP metrics are reviewed and discussed. In a case study in the U.S. Great Plains, surface phenological metrics of grasslands were extracted based on the Normalized Difference Vegetation Index (NDVI) time series derived from the 500 m Moderate Resolution Spectroradiometers (MODIS) eight-day composites. The remotely sensed Start-of-Season (SOS) and End-of-Season (EOS) were compared to the growing season indices that were calculated using the criterion provided by the United State Department of Agriculture (USDA), which defines the SOS and EOS as the last and first freezes (0°C) of the year. In addition, annual Growing Degree Days (GDD), a measure of heat accumulation to predict plant development rates, was calculated. The remote sensing-based SOS was further assessed by correlating with the accumulated GDD. Results showed various LSP metrics can be extracted from the NDVI time series effectively. The remote sensing-based LSP metrics and USDAdefined indices were unmatched. In our study site, the highest association between the remote sensingbased SOS and accumulative GDD was identified in warm-season short grass, followed by cold-season short grass, warm-season tall grass, and cold-season tall grass, in a descending order. When the discrete field phenological data were compared to the MODIS-based SOS of the grasslands, these data served as a good proxy.
# 5
Retzlaff, Rebecca • Molitor, Daniel • Behr, Marc • Wantzenrieder, Tom • Hoffmann, Lucien • (et. al.)
Abstract: Four soil management modalities (quadruplicate, 2010-2011) were assayed to determine their impacts on grapevine physiology and wine quality: Wolff mixture (1), natural greening (2), rotating harrow with winter greening (3) and natural greening with disturbance in dry conditions (4). The chlorophyll content was monitored monthly in 2011 and 2012 between June and September by optical measurement using the Dualex 4 (Force A, Paris, France) device. The field work in August was accomplished by a multi-angular flight campaign using a quadrocopter (md4-1000, microdrones GmbH) equipped with a six band VIS/NIR multispectral camera (MiniMCA-6). Bidirectional canopy reflectance was measured at three angles. In 2012, four flight campaigns were carried out throughout the grapevine season and compared to in-situ measurements of physiological factors. It was found that all four soil management modalities could be significantly discriminated using the MiniMCA-6 data. Discrimination was better for oblique viewing directions compared with nadir view, since the former configuration allows capturing a bigger canopy fraction in the image compared to nadir direction. Furthermore, leaf sampling and measurements for the calibration data set were not carried out in the upper leaf layer, which is almost exclusively visible in nadir view. A similar effect of the viewing directions was found for the quantification of chlorophyll content, and other leaf/canopy properties using stepwise multivariate regression: also here spectral data from the oblique directions led to most accurate regression models.
# 6
Eling, Christian • Klingbeil, Lasse • Kuhlmann, Heiner
Abstract: A georeferencing of the collected data is required for many unmanned aerial vehicle (UAV)-based remote sensing applications in the fields of surveying, precision farming, infrastructure inspection or geography. This georeferencing can generally be done indirectly using ground control points, or directly using an on-board sensor system. Since an indirect georeferencing is very time consuming and also not available in real time many users would prefer a direct georeferencing. However, UAVs do mostly only have a C/A code GPS and a low-cost IMU on board. Even if this sensor combination enables a rough navigation of the UAV it does not allow for a precise direct georeferencing. Thus, the development of a precise direct georeferencing system for lightweight unmanned aerial vehicles is currently in great demand. In this contribution a new developed direct georeferencing system for lightweight UAVs is presented, which is designed to (1) enable a precise position and attitude determination (position accuracy σ<5cm, attitude accuracy σ<1deg), (2) to be applicable on lightweight UAVs and (3) to be real-time capable for sampling rates >10 Hz. Generally, the system combines a dual-frequency GPS receiver, a tactical grade IMU, a single-frequency GPS receiver and a real-time processing unit on one board. Both, the RTK-GPS (real-time kinematic) and the attitude determination software are in-house developed and show good performances in first flight tests.
# 7
Neeland, Heiko • Kraft, Martin
Abstract: This work is about the mechanical construction and the measurement system of the ThuenoCopter. Both, the mechanical construction and the measurement system (sensors, 32 bit microcontroller board and image analyzing system) of the ThuenoCopter will be presented and first measurements of the low-cost image analyzing system will be shown. Climate changes and limited natural resources are the challenges for agriculture in a world where the population is still increasing and farm land may be decreasing for the next 50 years. On the one hand plant breeders need tools to measure many parameters like crop cover, crop temperature, etc. which are important for an optimal growth of plants. On the other hand improved conservative production methods for a sustainable agriculture can only be developed and tested practically if these parameters can be measured reliably and the relationship between these parameters and the plant growth is well known. The composition of the soil, the climate conditions, the supply with water and fertilizer and the use of plant protection chemicals determine the growth of every crop during its vegetation period. In order to have a UAV-based method for contactless crop inspection during the vegetation period - from tillage to harvesting - the ThuenoCopter has been developed on the base of an Oktokopter by Mikrokopter. The ThuenoCopter has a height of approx. 50 cm and its diameter is approx. 1 m. While flying along predefined routes, the measurement system of the ThuenoCopter measures the crop temperature, the air humidity and temperature, the global radiation, and it takes photos which are processed onboard in near real-time with the aim of detecting rows and weeds. The ThuenoCopter has been developed and optimized in weight and robustness (mechanical and electrical construction). A low-cost image analyzing system has been brought into operation. First measurements will be shown and discussed. Afterwards an outlook for improvements and further research activities is given.
# 8
Li, Fei • Miao, Yuxin • Chen, Xinping
Abstract: Timely and accurate qualification of aerial nitrogen uptake is of special significance to precision N management and recommendation for maize. Recent studies have confirmed the feasibility of retrieval of aerial N uptake of crops from spectral indices composed by the reflectance of 2-3 sensitive wavebands. In the present study, experiments involving different N rates in maize were conducted at Quzhou County of the North China Plain in 2009 and 2010. Several hyperspectral indices obtained from representative ratio- and area-based indices reported in the literature were selected to explore their potentials and stability for the estimation of aerial N uptake of maize across different growth stages, cultivars, sites and years. The results showed the optimum triangle vegetation index (OTVI) is most appropriate for aerial N uptake estimation with high correlation coefficients R2 of 0.84. Compared with triangular vegetation index (TVI), modified triangular vegetation index 1 (MTVI1) and modified triangular vegetation index 2 (MTVI2) with fixed bands, OTVI optimized by bands optimum algorithm increased R2 by 42%, 31% and 25%, respectively. The high correlation between the OTVI and aerial N uptake obtained in the different developmental stages of maize indicated that band optimized algorithms can potentially be implemented in future aerial N uptake monitoring by hyperspectral sensing.
# 9
Kooistra, Lammert • Suomalainen, Juha • Iqbal, Shahzad • Franke, Jappe • Wenting, Philip • (et. al.)
Abstract: To investigate the opportunities of unmanned aerial vehicles (UAV) in operational crop monitoring, we have developed a light-weight hyperspectral mapping system (<2 kg) suitable to be mounted on small UAVs. It is composed of an octocopter UAV-platform with a pushbroom hyperspectral mapping system consisting of a spectrograph, an industrial camera functioning as frame grabber, storage device, and computer, a separate INS and finally a photogrammetric camera. The system is able to produce georeferenced and georectified hyperspectral data cubes in the 450-950 nm spectral range at 10-100 cm resolution. The system is tested in an agronomic experiment for a potato crop on a 12 ha experimental field in the south of the Netherlands. In the experiment UAV-based hyperspectral images were acquired on a weekly basis together with field data on chlorophyll as indicator for the nitrogen situation of the crop and LAI as indicator for biomass status. Initially, the quality aspects of the developed light-weight hyperspectral mapping system will be presented with regard to its radiometric and geometric quality. Next we would like to present the relations between sensor derived spectral measurements and crop status variables for a time-series of measurements over the growing season. In addition, the spatial variation of crop characteristics within the field can be adopted for variable rate application of fertilizers within the field. The outcome of the experiments should guide the operational use of UAV based systems in precision agriculture systems.
# 10
Drauschke, Martin • Bartelsen, Jan • Reidelstuerz, Patrick
Abstract: In this paper, we describe two experiments regarding the monitoring of a test site in the Bavarian Forest National Park using unmanned aerial vehicles (UAVs) and we show their results. In the first experiment, we show that it is possible to relatively orient the RGB images acquired by a small UAV in power glider configuration without any flight stabilisation and without integrated navigation system (INS) initial values. This enables a 3D scene reconstruction, i.e., we obtain a point cloud showing distinctive 3D points. A much denser point cloud showing trees with branches can be derived from dense image matching. In the second experiment, we demonstrate how multispectral imagery can be interpreted on demand, i.e., without producing an ortho-mosaic, but using reliable features and a powerful classifier. With our algorithm, we follow up the aim to detect bark beetle attack in an early infection stage in Sitka spruce, Picea sitchensis, in the Bavarian Forest National Park.
spinning wheel Loading next page