Research On UAV Aerial Nursery Image Mosaic Algorithm Based On Feature Points

Image mosaie technology ean merge a group of original images with overlapping region into a whole image with high resolution and wide viewing angle through the procedure of registration and fusion,where no information of original images misses.In recent years,the nursery industry in our country has been transforming towards the direction of large scale and integration.Thus,the counting of nursery stock inventory is regarded as the application background of this paper.Automatic counting of nursery stock inventory can be implemented via taking the aerial photo of nursery area to be counted.However,the UAV cannot obtain the full image of large nursery area directly with the limitation of flight altitude and the camera focal length of UAV.To deal with the problem,the only solution is to merge the consecutive shooting nursery image sequence.The image mosaic algorithm based on feature points was chosen to be researched in this paper on the basis of the research background and signifieanee,the investigation of the research status at home and abroad,and the analysis of related algorithms.In this paper,the approach of the obtaining of UAV aerial nursery images was introduced;the image preprocessing of the collected aerial images was conducted;the image registration and image fusion,the committed steps of image mosaic,were researched emphatically,where the image registration approach was focused on the SIFT and Harris algorithm,and the image fusion approach was focused on the direct average method and fade-out image smoothing method;adaptive histogram equalization and RANSAC approaches were utilized to solve the problem of less number and mismatching of feature points.The eolleeted UAV aerial nursery images were chosen to conduct mosaic experiment,where the presented evaluation indexes with regard to the results of were compared and discussed.The experiment result proved that the number of matching feature points were increased and the problem of mismatching was solved with the improved algorithm,leading to a superior mosaic performance;better robustness in terms of UAV aerial nursery image mosaic co-uld be acquired with the method based on SIFT;the result of image fusion was better using the fade-out image smoothing method.At last,a software system on UAV aerial nursery image mosaic based on feature points was designed and developed on the basis of study in this paper,providing algorithm support,operating platform and system implementation for related practical application.