Narrow directional steerable filters in motion estimation
Computer Vision and Image Understanding
Computer Vision and Image Understanding
Ultrasound Image Sequence Registration and its Application for Thyroid Nodular Disease
Journal of Signal Processing Systems
Estimation of motions in color image sequences using hypercomplex fourier transforms
IEEE Transactions on Image Processing
| Hi-index | 0.01 |
Motion estimation in the spatiotemporal domain has been extensively studied and many methodologies have been proposed, which, however, cannot handle both time-varying and multiple motions. Extending previously published ideas, we present an efficient method for estimating multiple, linearly time-varying motions. It is shown that the estimation of accelerated motions is equivalent to the parameter estimation of superpositioned chirp signals. From this viewpoint, one can exploit established signal processing tools such as the chirp-Fourier transform. It is shown that accelerated motion results in energy concentration along planes in the 4-D space: spatial frequencies-temporal frequency-chirp rate. Using fuzzy c-planes clustering, we estimate the plane/motion parameters. The effectiveness of our method is verified on both synthetic as well as real sequences and its advantages are highlighted