基于灰度值相应概率的非刚性图像配准

提出了一种基于灰度值相应概率的，对多模态非刚性医学图像进行全局弹性配准的快速算法．通过两幅图像重叠部分的２Ｄ灰度联合直方图，计算灰度值相应概率（ＧＶＣＰ），使用遗传算法最优化ＧＶＣＰ实现两幅图像之间平移和旋转的全局刚体配准．将两幅图像的重叠部分均匀地划分成互为重叠的体积子块，再最大化每对对应体积子块图像灰度之间的ＧＶＣＰ，实现每对对应子块之间平移的局部刚体配准，并将每子块的中心作为一一对应的控制点．利用这些均匀分布的控制点对结合薄平板样条插值法实现图像的全局非刚性弹性配准．

Non-Rigid Registration for Multi-Modal Medical Images Based on Gray Value Correspondence Probability

Arapid algorithm to the non rigid registration ofm ulti m odalty im agesw asproposed using gray value correspondence probability (GVCP) com puted by 2D joint histogram as a registration criterion. Based on the global rigid registration (translation and rotation) by m axim izing GVCP betw een the intensity values ofvoerlapping partofCTand MRim ages, the overlapping partwas divided into uniform sub im age blocks overlapping w ith each other. The locally rigid registration (translation) was perform ed to produce a large am ount of control point pairs by m axim izing GVCP betw een the intensities of corresponding sub im ages. Agenetic algorithm wasused in the globaland localregistration optim ization. Finally, the globalelastic registration was perform ed com bined those corresponding points with thin plate spline.