Improving motion estimation by accounting for local image distortion

摘要: Cardiac elastography is a useful diagnostic technique for detection of heart function abnormalities, based on analysis echocardiograms. The the regional motion allows assessing extent myocardial ischemia and infarction. In this paper, new two-stage algorithm cardiac estimation proposed, where data taken from sequence 2D method combines advantages block-matching optical flow techniques. first stage employs standard (sum absolute differences) to provide displacement estimate with accuracy up one pixel. At second stage, corrected by estimating parameters local image transform within test window. are estimated in least-square sense. order account typical motions, like contraction/expansion, translation rotation, affine model assumed evaluated using 500 grayscale B-mode images, which generated as distorted, but known copies an original ROI, real echocardiogram. expressed terms errors: maximum error, root-mean-square average error deviation. errors proposed compared these cross-correlation interpolation sub-pixel space. Statistical shows that provides more accurate estimates than