Uav Remote Sensing Image Internal Distortion Correction Algorithm And Applied Research

In recent years, as a useful addition of Aerial remote sensing UAV remote sensing technology, remote sensing has a rapid development and wide range of applications. The face of serious geological disasters 5.12, due to steep topography of the affected areas, multiterminal flow changes, people can not achieve on-site investigation, to extract useful information, UAV, with its many advantages for the earthquake affected areas to assume the important task of image data. UAVs equipped with digital cameras as the sensor, secured the affected area topography image data at the first time, after a series of image processing technology, topography disaster recovery make for the earthquake relief command of the leadership to provide effective help . However, due to the difficulties of access Ground control points, the traditional method of geometric correction of remote sensing images can not be applied. Their must have a way to divide the cause of UAV remote sensing image into inside and outside the deformation of error for correction. This study will focus on the resulting image from the sensor to correct the internal error.First, several representative apply for UAV remote sensing image sensor, which digital camera calibration methods and shortcomings were be analyzed and summarized. It focuses on the only model with radial distortion algorithms and iterative algorithms such as the method of spaced, and the distortion factor to consider different combinations of integrated model and least square method, and the sub-region first-order polynomial correction, second and third order polynomial with the least square method. And then follow the camera calibration requirements for the laboratory. UAV remote sensing images taken using the same camera Canon 400D, As an experimental tool, took a high-resolution two-dimensional target image, and used Arcmap manually extract the target version of the image coordinates of the landmarks of the actual value, and press the camera image and the distortion rules calculate the ideal coordinates for computing needs. Correction of different model algorithm results were analyzed and compared in detail, from the x, y direction of the calibration error, point error, root mean square error, and the target image on any deviation from the two landmarks and other indicators suggest that: third-order polynomial model calibration results is the best, most accurate, and much better than other models. Third-order polynomial model was applied to test the film version of the camera calibration target image correction, the actual test results obtained the calibration accuracy of the results with the theoretical line. Finally, the third-order polynomial model in Matlab programming under the UAV remote sensing images on the film one by one for each pixel correction, and then using Spline interpolation function of the corrected image pixel gap left interpolation to good effect. That would be through an error correction UAV remote sensing images. Lsat take the stitching and random sampling and testing of a combination made the evaluation of calibration results.