Determining camera pose using free-form lines

Abstract

Determining pose or exterior orientation of a camera is an important step in surface reconstruction and subsequent steps such as perceptual organization and object recognition in photogrammetry or computer vision. Pose refers to position and attitude of a perspective camera at the time of exposure. It is generally determined using a set of points in object space and image space. Using free-form lines as a common feature between image and object spaces has always been a challenge. It is basically due to the unavailability of a mathematical model that relates free-form lines in object space, image space and pose of the camera. This paper introduces an innovative mathematical model to determine the camera pose using free-form lines. Everything recorded by a sensor can be considered a discretization of the physical surface [1]. Hence, it is less likely that the same points can be recorded or identified in both image and object space and it is usually easier to find corresponding lines. The method will lead a way to determine pose of images using historical Geographical Information System(GIS) or object space data.