Automated assembling of images: image montage preparation

Abstract

Image montage (or mosaic) finds a great deal of applications in ophthalmology, digital terrain mapping, and in autonomous navigation of re-entry vehicles. However, there is no direct method available for image montage preparation. Image registration methods may be used to solve this problem, but most of these methods are not useful, because they require either specific feature points in images or a great deal of overlapping between images of neighboring regions. A normalized correlation method is used for the preparation of montages when there is no rotational shift between neighboring images. A pyramidal approach is developed to reduce the computation. Further, a new method is developed for montage preparation when there is also a rotational shift between two neighboring regions. Different sets of subimages, which have both rotational and translational shifts with the subimages of neighboring regions, are taken for preparing image montages. The proposed method does not require any human supervision and the results are found to be very accurate.