Ground Based Agricultural Remote Sensing Platform On Energy Crop Biomass Monitoring

Agricultural remote sensing is a kind of ground information gathering technique, which includes computer science, database, network technique and geographic information system, global position system. Remote sensing is widely used in precision farming, such as crop nutrition evaluation, weed distribution, soil investigation, water resource investigation, vegetation recognition, crop yield prediction, insect infestation and agricultural environment monitoring.Miscanthus is considered a valuable candidate energy crop resulting from its high yield, the absence of known diseases, low water use, low nutritional requirements, and its non-invasive nature. Compared with traditional agricultural crop, energy crop has same features and its unique growth characters, so the agricultural remote sensing system monitoring energy crop can use the traditional or new method.Agricultural remote sensing platform includes satellite, aerial and ground platforms. Ground platform is used to measure wavelength and capture image near ground, such as tripod, tower, veicle and top of building. Aerial platform has aircraft, airship or balloon, but aircraft is more common and largely used. Space platform includes satellite, rocket, etc. Different area and resolution are derived on different height platform. These platforms can be used together within different situations. There are several important factors should be considered:spatial resolution, spectral resolution, radiometric resolution and time turnover.In order to monitor the energy crop biomass yield, the ground-based agricultural remote sensing platform was build; multispectral image geometric and radiometric rectification method was analyzed; image mosaic algorithm and energy crop biomass yield model was discussed. The details are listed as follow.Since current remote sensing method has drawback on energy crop monitoring, a stand-alone tower based remote sensing platform was build with high spatial resolution, high spectral resolution and low turn around time. Platform setup included system simulation, camera, pan tilt, tower and preset calibration.There are geometric errors resulting from equipments, weather, tower position, landform, etc. Radiometric errors are happened with solar elevation angle, sensor observation angle, electromagnetic radiation scattering and absorption by atmosphere, etc. Atmospheric correction is not considered on this ground-based remote sensing platform. Even under different lighting conditions, images were captured with automatic parameters to decrease the environment influence as much as possible everyday. Pinhole camera model were discussed to analyze the relationships between the extrinsic and intrinsic camera parameters and captured images location, which were used to get real world multispectral image without distortion.As the RS tower image mosaic method was different from traditional RS image, feature based image mosaic method was established after current image registration algorithm were discussed such as grey information based, transformation field based, etc. Image match is a procedure which gains the overlap areas and locations of two adjoining images under different time and angles, related to the image mosaic algorithm success and run time. Mosaic procedure in this project included feature extraction, registration based on feature, and mapping relationship between two images. Images on tower platform had so much obvious road information called line width information, which was used as feature to do mosaic except point feature, made the algorithm more accurate.Vegetation index was discussed to monitor the biomass yield. The relationship model between the dry matter production and vegetation index was build with accumulated index value, and harvest sample points biomass yield simulation was analyzed.The main innovation work was as follows:(1) The tower remote sensing platform was introduced to monitor energy crop biomass, and the radiation error was reduced by adjusting the cemera parameters using artificial neural networks.(2) The accuracy of processed images was improved through mosaic method based on feature and grayscale.(3) The accumulated vegetation index was constructed to simulate the sample point biomass production.