基于最近邻Marching Cubes的医学图像三维重建

3D reconstruction of medical images via nearest neighbor-based Marching Cubes algorithm

在医学图像三维可视化中,移动立方体算法(Marching Cubes,MC)是面绘制的经典算法。针对MC算法计算插值点导致执行速度慢、效率不高的缺点,提出一种基于最近邻逼近的MC算法,该方法在n次等分点量化序列中寻找等值面最近邻点代替线性或非线性插值,既避免了插值的大量计算又保证了误差精度,还可改善三角面片结构。利用可视化工具开发包VTK对人体脸部和脚部CT数据集进行三维重建,实验表明改进算法明显缩短了绘制时间,提高了重建效率。

Marching Cubes is a classical method for surface rendering in three-dimensional visualization for medical images. However, this algorithm implementation is slow and inefficient due to the linear or nonlinear interpolation operation. A new nearest neighbor-based MC algorithm is proposed to avoid the interpolation computation, which finds iso-points of the volume elements by nearest neighbor approximation to decrease the computational complexity and improve the quality of triangle faces. Experiments on human head and food CT data sets under the VTK toolkit demonstrate that the proposed method is competitive on visualization speed and rendering effect.