Optical flow estimation based on curvelet transform and spatio-temporal derivatives

Abstract EN

Optical flow estimation is still one of the key problems in computer vision.

When estimating the displacement field between two images, it is applied as soon as correspondences between pixels are needed.

The choice of the transform is an important tool in optical flow estimation.

So in this paper curve let transform is proposed to estimate the optical flow.

It is found that the proposed algorithm is much better than most recently previous work based on multi wavelet.

This is because curve let transform can be efficient in finding estimates along curves.