| CASI Aviation Hyperspectral Sensor is the internationally advanced and domestically leading airborne wide array hyperspectral imaging system.The sensor can build up to 288 continuous spectral channels with high spectral resolution in the spectrum of 380-1500 nm and has great utilization.The sensor has great value in such aspects as geological mineral exploration and development,ecological environment monitoring and protection,and urban development inspection and monitoring.When using the POS system(Position and Orientation System)to assist the geometric correction of CASI aviation hyperspectral data,there are two types of errors.The first is due to the fact that the phase center of the airborne GPS(Global Positioning System),the center of the inertial navigation system,and the center of the sensor cannot overlap,and the eccentricity vector caused by the deviation.The second is that the axis of the IMU inertial navigation system cannot be parallel to the sensor coordinate system,and the resulting alignment axis eccentric angle.These two errors will have an impact on the subsequent geometric correction correction accuracy.On the other hand,CASI hyperspectral imagery has a large amount of data.The traditional method is mainly based on general remote sensing images.It requires a lot of search calculations when solving the gray value of pixel points,and fails to make full use of computer high performance.This can lead to longer time processing for hyperspectral data processing.Therefore,it is very important to carry out high-precision and efficient geometric correction research.The data source of this paper is the CASI hyperspectral image of central Fujian Province.This paper studies the method of CASI aviation hyperspectral remote sensing geometric correction with POS system.The results of this paper are as follows:(1)In order to solve the eccentricity vector error and eccentricity angle error generated by CASI aviation hyperspectral remote sensing acquisition with POS system,this paper adopted the method of designing experimental field.For the two types of error generation principle,the appropriate solution model was selected to design the error compensation algorithm.After the error compensation,the average correction accuracy was improved by 1.580 meters,which reduced the effect of error.(2)The large amount of CASI hyperspectral data has an impact on the geometric correction processing time.In order to speed up the processing rate,a geometric correction method based on the block parallel distance search algorithm was proposed.First,the CASI image was divided into several blocks and then processed in parallel.It can make full use of the high computing performance of the computer.Secondly,for the problem that the calculation of the pixel gray value search was large,we designed the search box to perform the gray value search of the Pixel point.It controlled the range of contributing pixel points and reduced the amount of search calculations.It was verified by experiments that the fastest processing speed can be obtained when the number of threads used by the block parallel distance search algorithm was the same as the number of computer CPU cores.The geometric correction speed was 1.85 times higher than that of the bilinear interpolation at the optimal processing efficiency of 4 threads.(3)The flight attitude data recorded in the inertial navigation system was extracted during the aeronautical data acquisition process.The impact of large fluctuations in flight attitude was analyzed,and corresponding compensation measures were proposed.Improvement experiments were carried out by adding DEM data to reduce the impact of attitude data fluctuations.The average correction accuracy of the geometric correction was increased from 4.307 to 3.195 meters,which can achieve the desired accuracy requirements. |
PDF Full Text Request