Strip Of Uav Remote Sensing Image Stitching Algorithm Research

Unmanned Aerial Vehicle(UAV) has lots of unique advantages, which has been widely applied in many fields, especially in natural hazard investigation, etc. However, limited by UAV aeronautical shooting, the geographic area covered by a single image is small, it makes people can not receive enough imformation. There are a variety of unstable factors, such as weather, flight attitude, the lens limit and so on, all of which would influence the redution degree of the true surface features of the images. They format geometric distortion and radiation distortion, and reduce the stitching quality of two flight strip images, leading that establishing the panorama is very difficult, or even impossible. For leting UAV serve people better, it is needed to establish more appropriate model to solve this nodus.Firstly, this arctile describs the problems when the two UAV remote sensing strips images stitching, analyzes a few main reasons which produce the flight strip geometric error and deformation mode of these reasons, classifies these errors according to the mutual relationship of the accumulated, and proves by the experimental subjects:it is useless to improve the stitching quality of two flight strip images only through the traditional splicing in turn.Secondly, according to a characteristic that the accumulated error is greater when two flight strips stitching then when two single images stitching, and through simulation and comparison of the experimental data, this article uses the detecting method based on Sift algorithms and homography, compared to other algorithms the registration is higher when splicing flight strip images with them.On the basis of the above work, this article proposes center constraint to constrain the whole toward of single flight strip image, controls the deviating degree from the actual route; proposes symmetrical splicing method to reduce the accumulated error when splicing flight strip images; in order to further reduce the dislocation degree of the stitching seams, recommends LM algorithm to optimize the homography, for improving the local quality of the flight strip panoramic image; combines the advantages of center constraint, symmetrical splicing method and optimizing the homography, establishes the flight strip stitching model, effectively improve the quality of the flight strip mosaic.Finally, this arcticle uses a large number of experimental simulation, compares and evaluates the simulation results under different conditions, not only confirms the superiority of center constraint, symmetrical splicing method and optimizing the homography step-by-step, but also verifies that the flight strip splicing model posed by this arcticle is more effective then traditional order stitching method, through the realization, compare and evaluate of panoramic images.