Research on Low Altitude Aerial Image Stitching

In this paper, we study the problem of low altitude aerial image stitching. Though the traditional approach is well studied, batch-process image stitching is more difficult. In this work, we present a simple and effective approach to handle problems in low altitude aerial image stitching by considering GPS information of images and seam cutting algorithm. Our method develops an efficient approach to generate the flight path of UAV. We then select alternative images and calculate the overlapping region according to the distance between adjacent images. Features are efficiently detected not only in longitude overlap, but also in lateral, which largely improves stitching accuracy and decreases accumulative errors. In order to handle parallax caused by low altitude flight, we use seam finding method to avoid ghost. Experimental results show that our approach can effectively stitch a large number of images and successfully handle small parallax with low cost.