Dual bootstrap iterative closest point method and algorithm for image registration

A computer program product, comprising a computer usable medium having a computer readable program code embodied therein, wherein the computer readable program code comprises an algorithm adapted execute a method of performing image registration to map features in a first image to corresponding features in a second image in accordance with a transformation model that includes a parameter vector, said method comprising the steps of:

• establishing an initial bootstrap region as a current bootstrap region in the first image;
• estimating the parameter vector by minimizing an objective function with respect to the parameter vector, wherein the objective function depends on a loss function p(d/σ) of d/σ summed over selected features in the current bootstrap region, wherein d is a distance measure between q1 and q2, wherein q1 is a feature of the selected features after having been mapped into the second image by the transformation model, wherein q2 is a feature in the second image closest to q1 in accordance with the distance measure, and wherein σ is an error scale associated with the distance measures;
• calculating a covariance matrix of the estimate of the parameter vector;
• and
• testing for convergence of the iterative method, wherein upon convergence the current bootstrap region includes and exceeds the initial bootstrap region, wherein if the testing determines that the iterative method has not converged then performing a generating step followed by a repeating step, wherein the generating step comprises generating a next bootstrap region in the first image, wherein the next bootstrap region minimally includes the current bootstrap region and may exceed the current bootstrap region, wherein the repeating step comprises executing a next iteration that includes again executing the estimating, calculating, and testing steps, and wherein the next bootstrap region becomes the current bootstrap region in the next iteration.