Research On Aerial Image Mosaic Algorithm Based On UAV Images

UAV Remote Sensing technology emerged at the end of last century. This new type of remote sensing technology can get a large number of clear images from the low altitude. It has the characteristics of high flexibility, high efficiency, low cost and so on. It is gradually being widely concerned by people. UAV Remote Sensing is widely used in the safety inspection of dangerous area, geological environment and disaster assessment, the exploration of mineral resources and mapping and other related fields. But in the process of aerial photography, due to the influence of UAV flying height, camera angle of view and other parameters, the area covered by a single UAV image is not large enough. Therefore, in order to obtain more comprehensive information and expand the scope of the field of view in many specific tasks, need to splice multiple images, get a wide view of the panoramic view, the effective coverage of all work areas.At present there are many kinds of image mosaic method, among them image registration technology based on feature is robust to gray scale change, image rotation and illumination inequality. When the UAV getting the image, due to the different hardware equipment, light intensity, flight attitude and speed, it must cause the obtained Images to appear the uneven illumination, the rotation distortion, the scale distortion and so on. The image registration technology based on feature has been widely used in the field of the UAV Remote Sensing. The main research contents of this paper are as follows:Firstly, Introduces the research background and the research status quo of UAV aerial image mosaic, and the basic theoretical knowledge of image mosaic is briefly described. The characteristics and common methods of UAV image mosaic are analyzed in detail.Secondly, in this paper, the SIFT algorithm is used to extract the feature of UAV image. Because the UAV image has the characteristics of high quantity and high resolution, when extracting features by the SIFT algorithm, will appear the situation of Image mosaic overall time greatly increase due to the feature point for preliminary extraction too much. To solve this problem, this paper improves the search strategy of the extreme point, reasonably reduces the number of feature points, so it can effectively reduce the time of SIFT algorithm splicing.Thirdly, in this paper, the RANSAC algorithm is used to eliminate the error matching point pair. Aiming at the problem of too many iterations of the RANSAC algorithm due to the high proportion of false matching points, the RANSAC algorithm is improved and a Hough-RANSAC algorithm is proposed. The experimental results show that under the premise of guaranteeing the image mosaic effect, this algorithm improves the operation speed of the algorithm when the false matching rate is higher.