Sparse Based Optical Flow Estimation in Cardiac Magnetic Resonance Images

摘要

Optical flow enables the accurate estimation of cardiac motion. In this research, a sparse based algorithm is used to estimate the optical flow in cardiac magnetic resonance images. The dense optical flow field is represented using a discrete cosine basis dictionary aiming at a sparse representation. Optical flow is estimated in this transformed space by solving a L1 problem inspired on compressive sensing techniques. The algorithm is validated using four synthetic image sequences whose velocity field is known. A comparison is performed with respect to the Horn & Schunck and the Lucas and Kanade algorithm. Then, the technique is applied to a magnetic resonance image sequence. Results show an average magnitude error as low as 0.35 % for the synthetic sequences, however, results on real data are not consistent with respect to the obtained by other algorithms. This fact suggests the need for additional constrains to cope with the dense noise.