Medical Image Registration Based on Inertia Matrix and Iterative Closest Point

Abstract

The closest iterative point algorithm (ICP) is widely used in medical image registration. But there exist some problems in the following aspects. First, due to its heavily computational load, it has a time-consuming process and a low registration efficiency. Second, due to the fact that it heavily depends on whether the initial rotation and translation matrices of the floating point set can be exactly extracted, it often traps in the local optimum and even fails to register images. In addition, due to the complexity of medical images, it is difficult to automatically extract the salutary feature points. In this paper, by computing the coordinate inertia matrices of the reference and floating images, the rotation angles are obtained and referred to as the initial rotation parameters of ICP for image registration. The edges of the reference and floating images are detected by the edge convolution kernel so-called B-spline gradient operator (BSGO) and then the binarization images involving the feature points are acquired. The experimental results reveal that, this proposed method has a fairly simple implementation, a low computational load, a fast registration and good alignment accuracy. Also, It can efficiently avoid trapping in the local optimum and cater to both mono-modality and multi-modality image registrations.