A new calibration algorithm using only angles is presented.
The proposed algorithm is based on the simple idea of the
invariance of angles under the similarity transformation.
Stratified calibration approaches using various scene
constraints including angles have been proposed. However,
the proposed algorithm uses only one type of scene
constraint of angle, while directly recovering Euclidean
structure from projective structure. We demonstrate the
feasibility of the proposed algorithm through the
experiment using synthetic and real images.