RECONSTRUCTION FROM IMAGE SEQUENCES BY MEANS OF RELATIVE DEPTHS

This paper deals with the problem of reconstructing the locations of n points in space from m different images without camera calibration. I t shows how these problems can be put into a similar theoretical frame work. A new concept, the reduced fundamental matrix, is introduced. It contains just 4 parameters and can be used to predict locations of po ints in the images and to make reconstruction. We also introduce the c oncept of reduced fundamental tensor, which describes the relations be tween points in 3 images. It has 15 components and depends on 9 parame ters. Necessary and sufficient conditions for a tensor to be a reduced fundamental tensor are derived. This framework can be generalised to a sequence of images. The dependencies between the different represent ations are investigated. Furthermore a canonical form of the camera ma trices in a sequence are presented.