Research on Image-based Airline Correction Xiong Zhen, Wang Xiangjun, Zheng Lanfen, Tong Qingxi (Institute of Remote Sensing Application, Chinese Academy of Sciences, Beijing 100101) The data used for airborne image correction is to use airborne data to calculate various attitude parameters of aircraft flight The remote sensing image obtained by the airborne imaging spectrometer is subjected to azimuth correction, so that the remote sensing image reaches the result of geometric coarse correction, which lays the foundation for geometric fine correction. OM IS is a 128-channel imaging spectrometer newly developed by Shanghai Institute of Technical Physics, Chinese Academy of Sciences, and conducted a test flight in Changzhou, Jiangsu, in August 1998. Combined with this experiment, the airborne GPS data was used to conduct flight path correction research, and the actual flight data was used to conduct the experiment, and good results were obtained.
1 Introduction Aviation remote sensing technology is widely used in agriculture, forestry, geology, petroleum, surveying and mapping, urban planning and other fields. However, because the aerial remote sensing platform, the flying height of the aircraft is relatively low, and is greatly affected by the airflow, it is quite difficult to maintain the flight straight. Therefore, the geometric quality of the aerial remote sensing image is obviously inferior to that of the satellite remote sensing image. Because the satellite's flight orbit and flight attitude are known, these off-the-shelf parameters can be used for geometric correction of satellite remote sensing images. For aerial remote sensing images, geometrical correction is much more difficult due to unknown flight parameters. Therefore, the geometric correction of aerial remote sensing images is a very important subject. In the past, the geometric correction of remote sensing images was mainly carried out through ground control points. The use of airborne GPS data for route correction is a new attempt in the geometric correction of aviation remote sensing images.
The method and steps of data used for route calibration 2.1 Reading GPS data files Reading GPS data files should solve two problems, one is coordinate conversion and the other is time delay processing.
2.1.1 Coordinate conversion Since the recorded coordinates are all geodetic coordinates, the geodetic coordinates must be converted into rectangular coordinates of Gaussian plane before use. In order to ensure that the image quality is not compromised by coordinate conversion, a more rigorous coordinate conversion formula is adopted. At the same time, in order to minimize the geometric distortion caused by coordinate conversion, the average meridian of the area where the image is located is selected as the projected central meridian during coordinate conversion. The formula for converting from geodetic coordinates to Gaussian coordinates is as follows: B is latitude and l is longitude difference.
2.1.2 Time delay processing Generally, the receiver updates the data once every second, and the number of records per second is 15. Due to hardware reasons, there may be a time lag in data recording. After research, it is found that this lag generally does not exceed 3 records, that is, no more than one-fifth of a second, and this lag does not necessarily occur in all records, sometimes there is only one remote sensing technology and application time delay, sometimes there are several The item has a time delay at the same time. As shown in Table 1, when the time changes in the fourth line, the GPS data record items have a time delay in altitude and speed, and their lag time is one fifteenth of a second. All other items have jumps, no Time delay occurs. Because this kind of time delay phenomenon is very common, in order to avoid the impact of time delay, the data must be processed. Considering that the time delay does not exceed one-fifth of a second, during data sampling and recording, three records after the time jump are selected for comparison at the same time, and the latest record among the three records shall prevail.
Line year, month, day, time, latitude, longitude, altitude, linear velocity 2.2. Calculation of flight attitude GPS data generally includes such items as: time, longitude, latitude, altitude, and linear velocity. Based on these data, the flight path of the aircraft can be fitted in the horizontal and vertical planes, and then the attitude parameters can be calculated. There are three attitude parameters calculated here: pitch angle j, course azimuth angle κ, roll angle k.
2. 2. 1 Pitch angle Fit a curve according to the height data in the vertical plane, and then calculate the flight pitch angle at each updated data point (ie node, hereinafter referred to as node) according to the adjacent height data. Since the photography stabilization platform was not used during the flight experiment, the aircraft's pitch angle can be considered as the scanner's pitch angle, thereby correcting the image.
Among them, s is the distance between the two points before and after the node, s = ((x is the Gaussian coordinates of the node, H is the altitude of the two points before and after the node, j is the elevation angle, j 0 is the elevation angle, j 0 is the depression angle.
2. 2. 2 The azimuth of the route fits a curve according to the data in the horizontal plane, and then calculates the flight azimuth at each node based on the adjacent coordinate data.
Among them, Δx, Δy are the vertical and horizontal coordinate difference of two points before and after the node.
2. 2. 3 roll angle is in the horizontal plane, according to the curve fitted by GPS data, calculate the radius of curvature r at each node, and then calculate the roll angle of the aircraft according to the principle of dynamics. Similarly, because the photography stabilization platform was not used during the flight experiment, the rollover angle of the aircraft can be regarded as the rollover angle of the scanner to correct the image.
The radius of curvature r is estimated as follows: Among them, the coordinate x of the center of the circle can be obtained by solving the system of linear equations, which is the coordinate of the midpoint of the vertical bisector of the line segment.
The force situation of the airplane is shown in Figure 1, where centripetal force F is where, m is the mass of the airplane, v is the linear velocity of the airplane, and r is the radius of curvature of the flight trajectory.
The aircraft's inclination angle, or roll angle k, is: where g is the gravitational constant.
2.3 Calculate the attitude of each scan line After solving the attitude parameters at each node, you can calculate the attitude parameters of each scan line. The posture parameters of each scan line can be interpolated according to the posture parameters of each node or the curve fitting data can be used to interpolate the coordinate data of each scan line, and then the posture parameters of each scan line are calculated. This article uses the former algorithm, the interpolation algorithm (Figure 2).
2.4 Calculate the coordinates of each pixel First calculate the coordinates of the middle point of each scan line, and then calculate the coordinates of each pixel according to the instantaneous field angle IFOV of the scanner. The spatial coordinates of the main point of the image before rotation transformation are (X, Y , H), the coordinate after rotation transformation is (X where, a here, ψ is the heading angle, that is, the pitch angle, ω is the lateral angle, that is, the roll angle, and κ is the azimuth of the route, as shown in Figure 3.
The formula for calculating pixel coordinates is as follows: where T is the azimuth of the scanning direction, k is the roll angle of the center point of the scan line, k is the roll angle of the left end of the scan line, and k is the roll angle of the i point, i = 1, 2,. .., 512 h is the distance from the center point of the ground scan line to the scan center point, H is the altitude, s is the distance from the i-th scan point to the center point of the scan line, x 1, y 1 are the coordinates of the center point of the scan line, x is the coordinate of the first scan point.
2. 5 resampling Calculate the position after sampling according to the pixel coordinates, and then assign the gray value of the pixel to each adjacent pixel, where the reciprocal of the distance is the weight, Q (u, v) is the output image The DN value of the pixel, P is the result of the surrounding pixel 3. This experiment uses the data of the 128-channel imaging spectrometer OM IS developed by the Shanghai Institute of Technical Physics, Chinese Academy of Sciences in August 1998 in Changzhou, Jiangsu, a total of 8 554 Scanning lines, each scanning line has 512 pixels, and the instantaneous field angle IFOV is 3 milliradians.
The images before and after the course correction are shown in Figures 5-9.
4 Discuss the image after the route correction. Figure 6 is the track graph drawn by GPS data. Comparing Figure 6 and Figure 7, it can be seen that the airborne GPS data can be used to correct the yaw problem of the image. The corrected image is completely faithful to the actual trajectory of flight. FIG. 8 is a part of the original image without course correction, and FIG. 9 is a part of the image after course correction. Both images are taken from the beginning of the image. As can be seen from Figure 9, the image still has geometric distortion. Analysis of this deformation shows that it is no longer caused by the distortion of the route, but by the high-frequency jitter of the scanner. If a photographic stabilization platform is used during the flight experiment, this geometric deformation will disappear. At present, the Shanghai Institute of Technical Physics of the Chinese Academy of Sciences is developing this stable platform for aerial scanning. It is believed that the problem of geometric distortion of the image due to high-frequency jitter of the scanner will be solved soon. In addition, as can be seen from Figure 9, the remote sensing image after route correction still has image edge distortion and height difference distortion, which shows that this route correction only solves the yaw problem of the image, and cannot solve the geometric correction. For all problems, the auxiliary data such as ground control points must be used for further geometric correction.
5 Conclusion (1) Using the onboard GPS data to calculate the flight attitude parameters of the aircraft, it is possible to perform course correction well.
(2) There are still some other geometric distortions in the aerial remote sensing image after the course correction, such as edge distortion and height distortion of the image. If there is no photographic stabilization platform while flying, there is still geometric distortion caused by the high-frequency shake of the scanner after the route is corrected.
[1] Japan Remote Sensing Research Association. Remote Sensing Explained [M]. Beijing: Surveying and Mapping Press, 1993. 174 [2] Guo Defang. Computer processing and pattern recognition of remote sensing images [M]. Beijing: Electronic Engineering [3] Zhu Zhaoguang, Sun Hu, Cui Bingguang. Photogrammetry [M]. Beijing: Surveying and Mapping Press, [4] Kong Xiangyuan, Mei Yiyi. Control Surveying [M]. Beijing: Surveying and Mapping Press, 1991. 266 [5] Cheng Shouzhu, Jiang Zhiyong. General Physics [M]. Beijing: People's Education Press, [6] Zheng Wei, stated Peng. Outline of Resource Remote Sensing [M]. Beijing: China Science and Technology Press,
Jumei Video(Shenzhen)Co.,Ltd , https://www.jmsxdisplay.com