基于交叉累计剩余熵的图像配准中插值方法的改进



交叉累计剩余熵（CCRE）比传统互信息在配准强噪声图像时更具优势,但采用部分体积（PV）插值的CCRE在网格点容易产生局部极值,不利于变换参数的优化。针对该问题,研究基于3阶B样条函数的PV插值（BPV）、哈宁窗sinc函数的PV插值（HPV）和Blackman-Harris窗sinc函数的PV插值（BHPV）方法在CCRE中的应用,提出一种新的插值方法。该方法采用灵活的邻域中心,将插值点对联合直方图贡献的权重分散到临近的9个点上,并使用高斯函数作为PV插值的核函数,避免权重突变。实验结果表明,与BPV,HPV和BHPV插值方法相比,

-Improved cross cumulative residual entropy (CCRE) image registration interpolation method based on cross cumulative residual entropy of information advantage when strong noise image registration than traditional mutual, but with partial volume (PV) grid point interpolation CCRE prone to local extreme, is not conducive to transformation parameter optimization. To solve this problem, based on 3-order B-spline interpolation function PV (BPV), Hanning window sinc interpolation function PV (HPV) and Blackman-Harris window sinc interpolation function PV (BHPV) Application of CCRE, a new interpolation method. The method uses a flexible center of the neighborhood, the interpolation point joint histogram contribution weights dispersed to nine points nearby and use the Gaussian function as a PV interpolation kernel function, avoid weight changes. Experimental results show that compared with BPV, HPV and BHPV interpolation method,