| چکیده انگلیسی مقاله |
Introduction Todays, the application of new methods for monitoring and online management of agricultural farms is necessary to increase quality and quantity of agricultural products. Remote sensing is one of the technologies that can be used to monitor agricultural farms and natural resources. This technology is capable of detection and prediction of farm changes by application of satellite and aerial images. Satellite images have some limitations for application in agriculture, namely, high cost, low revisit time and low spatial and spectral resolutions. Thus, recently design of low altitude remote sensing system is considered as a useful tool for ground observations with high spatial resolutions. This system is used to monitor and collect online images for agricultural farms of small spatial farms. So, by the importance of aerial imaging in agricultural farm management and limitations of satellite imagery, the main objective of this research is to develop an unmanned aerial remote sensing system for imaging agricultural farms with high spectral and spatial resolutions. Materials and Methods The constructed unmanned aerial vehicle (UAV) is composed of aerial part and ground station. Aerial part is composed of carbon fiber body and arms, 8 brushless DC electric motors, 8 control speeds, control board, PID controller, AHRS system for roll, pitch and yaw angles measurements (with 3 gyroscopes, 3 accelerometers, 3 Magnetometers, one barometer), GPS, camera mount with 2 servo motor for compensation of camera vibration, 3-cell chargeable Li-Po battery (5000mA, 11.1 V) and voltage measurement unit. A multispectral camera (ADC-Micro, Tetracam Company) with 520-920 nm wavelengths and 3.2 Megapixel CMOS sensor in Green, Red and Near Infra Red bands (6-12 V, 2 Gb memory). Ground station is constructed for conduction and control of aerial part and, indeed, it is connected with operator to aerial part. Ground station is composed of 8 frequency radio control (2.4 GH frequencies), flight monitor software and control software. Total system weight is 2 kg with 78 cm length and width and 29 cm height. The main duty of ground station is following current position and coordination of the UAV, visiting flight parameters and adjustment of primary flight parameters. Sending online imagery of regain to the ground station is carried out by a CCD camera. To connect aerial and ground parts together, a telemetry system is used. To evaluate the performance of the system, collecting data is carried out in a wheat farm in the Mohammad Shahr, Karaj, Iran. Results and Discussion Some factors such as flight endurance, flight maximum preload, maximum and minimum UAV speed, flight altitude, camera mount performance and spatial resolution are investigated. So, imagery are processed, NDVI and supervised classification map is extracted. To evaluate the results of classification, error matrix and overall accuracy is calculated. Based on the results, manuaribilty and stability of the UAV during the flight and also in takeoff and landing positions were satisfactory. The results of experiments showed that the average time endurance of the UAV with installation of GPS, CCD camera, Multispectral camera is 10 minutes. Thus, the results of UAV preload experiments indicated that the maximum preload of the developed UAV is 1 kg. Also, the maximum wind speed for the UAV flight based on the result of this research was 15km/hr. The performance of the system is decreased by increasing wind speed. The optimum UAV speed for collecting qualified imagery is 0-40 km/hr. In higher speed, the quality of imagery will be decreased. Conclusion The obtained results from this study are: - Spatial resolution: based on the results of experiments in the height of 250 m by image dimension (2048*1536), the total covered area in imagery is 2.8 ha. So, time resolution based on the UAV speed and imagery saving file format is 2-5s. So, spatial resolution of imagery in the height of 10m was 3.6 mm/pixel and in the height of 250m the obtained image is 95mm/pixel. - NDVI map showed higher value in denser regains. - Supervised classification map (maximum likelihood): the results of experiments showed that the overall accuracy of classification map was 93.99% and Kappa coefficient obtained 0.9. Classification results showed that different classes such as soil, crop canopy, shadow and etc. were separated and recognized completely. High value of the overall accuracy and Kappa coefficient indicates that the UAV is capable of taking good imagery without nose which shows the satisfied performance of control system of the UAV and control system of camera mount in order to keep the UAV in the right situation and to fix camera in the desired position. |