Development And Application Of Uav Thermal Infrared Image Acquisition System

The UAV thermal infrared remote sensing makes the content of agricultural information extraction more abundant with its unique advantages in terms of spectral information,which can greatly improve the management level and efficiency of farmland in China.Canopy temperature and crop water status are closely related,thus monitoring of canopy temperature is one of the important means to improve the yield and quality of crops.Therefore,it is important to study on the UAV infrared image acquisition system and the model of crop canopy temperature.Due to the problems that the thermal infrared camera can only capture timing images,no POS data and that thermal infrared data acquisition have not realized by flight controller.In this paper,we develop a UAV airborne thermal infrared image acquisition system,which can obtain stable and stitchable thermal infrared remote sensing image,based on STM32 embedded development technology.The main contents of this paper are as follows:1.We completed the development of airborne thermal infrared camera platform,to meet the requirement of thermal infrared remote sensing data acquisition stability.By analyzing the demands of thermal infrared image acquisition,we set the main design parameters of six-rotor UAV,selected the UAV control system and finished UAV assembly and debugging,to ensure the flexibility and stability.We used a multi-rotor performance evaluation software xcopterCalc and the practical flight test to verify the rationality of six-rotor UAV design.It turn out that the six-rotor UAV met the thermal infrared camera carrying requirements.2.On account of STM32 embedded development technology,we accomplished the development of thermal infrared image acquisition module.According to the module design scheme,we chose STM32F407 as the main control chip to complete the system hardware design,including the design of the main control module,memory module and interface module.On the basis of module hardware design,we carried out program design and development,which included the overall scheme of software system design and functional modules program design.Through board level testing and system testing,the module performance was validated at last.The test results show that the module developed in this paper can satisfy the design requests.3.By means of combining the study of Pixhawk control signal format and FLIR Tau 2 camera shutter trigger principle,we put forward thermal infrared data acquisition flight control method.Since the existing POS data acquisition tool cannot recognize the thermal infrared image format,we put forward the method for POS data extraction based upon the analysis of related Pixhawk parameters and flight log.By generating thermal infrared orthophoto map,we anaysed the accuracy of POS data.Image mosaic results show that the accuracy of the POS data met stitching needs.4.To ensure the best system data acquisition capability,we put thermal infrared camera calibration and correction into practice.In this paper,we calibrated the camera using Matlab calibration toolbox based on the principle of chessboard.We implemented non-uniformity correction via the method of two-point temperature correction,reducing the IRFPA non-uniformity to 1.27%.The UAV flight parameters and camera parameters are closely related,we set the UAV flight speed,flight altitude and flight trajectory parameters reasonably to guarantee the image stitching quality and optimal trajectory planning.5.We put the UAV infrared image acquisition system developed into application by researching on the model of crop canopy temperature.After collecting maize mature period thermal infrared images,we made use of image mosaic software Pix4 D and remote sensing software ENVI to process data.We established a linear regression model between the thermal infrared image gray value and canopy temperature.The linear regression equation is(?)=0.0407-277.68,the determination coefficient is 0.9462 and the standard error of estimate equals to 0.73.The results have shown that the model has a high level of goodness of fit,can be applied to crop water stress monitoring.