Tenth IEEE International Conference on Computer Vision (ICCV'05) Volume 1
DOI: 10.1109/iccv.2005.33
Full text: Download
The invariance of the similarity measure in photomet- ric distortions as well as its capability in producing sub- pixel accuracy are two desired and often required features in most stereo vision applications. In this paper we pro- pose a new correlation-based measure which incorporates both mentioned requirements. Specifically, by using an ap- propriate interpolation scheme in the candidate windows of the matching image, and using the classical zero mean normalized cross correlation function, we introduce a suit- able measure. Although the proposed measure is a non- linear function of the sub-pixel displacement parameter, its maximization results in a closed form solution, resulting in reduced complexity for its use in matching techniques. Ap- plication of the proposed measure in a number of bench- mark stereo pair images reveals its superiority over existing correlation-based techniques used for sub-pixel accuracy.