Image alignment and stitching: a tutorial
Foundations and Trends® in Computer Graphics and Vision
Image stitching with dynamic elements
Image and Vision Computing
Graph-Based global optimization for the registration of a set of images
PSIVT'06 Proceedings of the First Pacific Rim conference on Advances in Image and Video Technology
| Hi-index | 0.00 |
This paper presents a system for creating a full 360-degree panorama from rectilinear images captured froma single nodal position. The solution to the problemis divided into three steps. The first step registers alloverlapping images projectively. A combination of agradient-based optimization method and a correlationbased linear search is found to be robust even in casesof drastic exposure differences and small amount of parallax. The second step takes the projective matrices andtheir associated hessian matrices as inputs, and calibrates the internal and external parameters of every images through a global optimization. The objective is tominimize the overall image discrepancies in all overlapregions while converting projective matrices into cameraparameters such as focal length, aspect ratio, image center, 3D orientation, etc. The third step re-projects allimages onto a panorama by a Laplacian-pyramid-basedblending. The purpose of blending is to provide a smoothtransition between images and eliminate small residuesof misalignments resulting from parallax or imperfectpairwise registrations. The blending masks are generated automatically through the grassfire transform. Atthe end, we briefly explain the necessary human interface for initialization, feedback and manual options.