Robust Scale Iterative Closest Point Algorithm Based On Correntropy For Point Set Registration

Point set registration is important for calibration of multiple cameras, 3D reconstruction and recognition, etc. The traditional scale iterative closest point (ICP) algorithm is fast and accurate for scale registration of point sets, but it performs worse when the point sets with large outliers or noise. This paper introduces a novel algorithm based on correntropy for scale registration of noisy point sets. Firstly, the definition of correntropy is introduced and its property of eliminating outliers is demonstrated by comparing with the mean square error (MSE). Secondly, a novel objective function of scale registration problem is proposed by applying the maximum correntropy criterion (MCC). After that, a new scale ICP algorithm is proposed to solve this energy function. This method uses a simple iterative algorithm and computes the scale transformation by methods including half-quadratic technique and derivation quickly at each iterative step. Similar to the ICP algorithm, this new algorithm converges monotonically to a local maximum for any given initial parameters. Experimental results demonstrate that our algorithm has the high speed and accuracy for scale registration with outliers compared with the traditional scale ICP algorithm and the state-of-the-art algorithms.