Panoramic Mosaic of Aerial Image of Drone based on L-ORB
Author:Zhiming Chen, Chunxue Wu, Yan Wu Author Unit: School of Optical-Electrical and Computer Engineering, University of Shanghai for Science and Technology, China,200093 O’Neill School of Public and Environmental Affairs, Indiana University, Bloomington，USA.47405 Department of Computer Science and Technology, Tianjin University
Abstract：In order to meet the real-time and robust requirements of UAV aerial image stitching, this paper proposed L-ORB algorithm based on the improved FAST feature point extraction and description (ORB) algorithm, and combined with the approximate nearest neighbour search (ANN) and the improved sample consistency (PROSAC) algorithm to complete UAV aerial image stitching. Firstly, through the parameters of UAV cameras to calculate the overlapping area of the pictures, to obtain the distribution of feature points for reducing the detection area. Without affecting the accuracy of feature point matching, using the improved FAST feature point algorithm to search feature points on specific ROI domain and generate BRIEF feature description factors. Then use approximate nearest neighbour search (ANN) and local sensitive Hash (multi-probe LSH) to speed up the search of feature points that will be matched. Finally, PROSAC algorithm is used to eliminate false matches. Experimental results show that the algorithm can reach a high balance in accuracy and speed.
Keywords：UAV aerial image stitching;L-ORB
Download information [SIZE：10.60 MB Amount of downloads： second]
