Custom aspect ratio correction for unwrapped omnidirectional view images

The growing utilisation of omnidirectional view cameras in robotic applications is mainly owing to their wide 360° field of view. This paper addresses the issue of incorrect aspect ratio often found in cuboid panoramic unwrapped spherical omnidirectional view images. The proposed method consists of an efficient computational technique that utilises only three planar image points obtained from the omnidirectional view's setting. The correction of the aspect ratio in turn improves the matching performance of Scale-Invariant Feature Transform keypoints in the unwrapped omnidirectional view images. Experimental results are presented towards the end of this paper as an empirical verification of our proposed method.