Region-growing motion tracking method based on bayesian inference and polynomial fitting

A region-growing motion tracking method based on Bayesian inference and polynomial fitting, comprising the following steps:

• Step 1:
• determining a seed point:
• selecting a region of interest (ROI), and then selecting a candidate point set from displacement estimation points in the ROI;
• calculating posterior probability values of a candidate point in the candidate point set and four neighborhood points, a sum of maximum posterior probability values (MPPS) of these five points, and a maximum absolute displacement difference;
• judging whether the candidate point in the ROI is a seed point according to a set threshold, the MPPS and the maximum absolute displacement difference;
• Step 2:
• performing guided search, displacement estimation:
• determining an initial known-displacement point set, an interior point (IP) set and a boundary point (BP) set after the selection of the seed point, selecting a point with the maximum posterior probability value as an active growing point within the determined BP set, then conducting guided search according to a displacement of the active growing point, and determining a neighborhood and a search range for displacement estimation according to a location and the displacement of the active growing point;
• Step 3:
• conducting local polynomial fitting:
• when the displacement difference between the displacement estimation point and its neighborhood point is higher than the previously set threshold, the local polynomial fitting is carried out with the displacement estimation point as a center, and the local polynomial fitting is carried out using the IP point around the displacement estimation point, and a fitting value is taken as a new displacement estimation value;
• Step 4:
• calculating strain elastogram:
• obtaining an ultrasonic strain elastic image through a differential calculation of displacement field of ultrasonic elastic motion.