Hyperspectral Imaging System Optimization and Image Processing

In the research of remote sensing for precision farming applications, the data quality of the aerial hyperspectral imaging system suffers from geometric distortions. Some of the distortions are caused by aircraft attitude change during the current pushbroom type image scanning. These distortions must be corrected before image data can be geo-referenced and used for field pattern identifications. Development of methods and algorithms for the correction of this type of remote sensing data distortion is the objective of this study. Three different approaches, namely manual correction, sensor augmentation, and image processing were developed. A Fiber Optic Gyroscope (FOG) attitude sensor was used on board the airplane to measure the real-time image sensor attitude, a polynomial interpolation algorithm was developed, and a reference straight feature was segmented based on the selected training dataset. The performance of all three methods was evaluated and compared. It is suggested that further integration of the attitude sensor with the imaging system can provide an instantaneous fully automated distortion correction system.