iat_ransac

[ INLIERS, MODEL ] = iat_ransac( PTS1, PTS2, PROB, TRANSFORM, MAXITER, MAXINVALIDCOUNT, TOL )

iat_ransac implements the ransac algorithm adapted to estimate the optimum TRANSFORM from point correspondences PTS1<->PTS2 and returns the largest inlier set of correspondences and the respective transformation. Note that it uses a symmetric error function to choose the inliers.

Input arguments

PTS1 A 3xN array that holds the initial set of homogeneous coordinates
PTS2 A 3xN array that holds the set of transformed homogeneous coordinates (in correspondence with PTS1)
TRANSFORM The type of the transform, valid strings: {‘homography’, ‘affine’, ‘similarity’, ‘euclidean’, ‘translation’}

 

Optionally, the user can define his own parameter values by

[ INLIERS, MODEL ] = iat_ransac( PTS1, PTS2, TRANSFORM, ‘PARAM1′, PARAM1VALUE,…)

Optional Input Parameters

‘maxIter’ The maximum number of iterations allowed. You may need more iterations if the result is not sufficient (default: 100)
‘prob’ The desired probability to pick a MSS(Minimum Sample Set) with no outliers (default: 0.99)
‘maxInvalidCount’ The maximum number of degenerated set picks (default:100)
‘tol’ The tolerance for the symmetric error that determines whether a data point is an inlier or
an outlier. Since the coordinates of points are normalized before the RANSAC core steps, the error tolerance can be lower than one (default: 0.5)

 

Output arguments

INLIERS The indices of points in PTS1 (or PTS2) that belong to the largest inlier set
MODEL The RANSAC-optimum estimated transform. It corresponds to the output of the function iat_get_<TRANSFORM> when using the inlier set only.

 

Back to Index