Multi-aperture optical flow estimation for artificial compound eye using non-subsampled contourlet transform and guided filter

Abstract

Abstract A multi-scale optical flow estimation for the image captured by artificial compound eye (ACE) is investigated in this article. The optical flow estimation of ACE must be adapted by designing algorithms according to its unique multi-aperture characteristics. A more general filter for the regularization term, rather than a single iterative solution in the traditional variational model, is devised using the non-subsampled contourlet transform to enforce band decomposition and estimate the optical flow field. To circumvent the spillover and error of the single-aperture fringe flow field, a flow gradient weight is introduced to suppress it and enhance motion details. Furthermore, low-pass subbands adopt the Bayes threshold with the advantage of efficiently eliminating outliers. More high-pass subbands adopt guided filter with the benefit of separating important details from outliers. The prominent feature of the proposed method is that the accuracy of optical flow estimation is improved effectively by eliminating outliers. Finally, experimental results demonstrate the superiority of the examined optical flow estimation.