基于局部平面信息的三维灰度图像方向内插方法

A Directional Interpolation Method for 3 D Gray Scale Image Based on Local Plane Information

灰度图像插值是体绘制的预处理步骤。插值结果的好坏， 直接影响着绘制的效果。通常， 插值错误主要出现在组织的边界处。组织的表面上灰度的变化较小， 如果在表面上做插值， 就可以保证插值在相同组织之间进行，避免了常用的线性插值会产生新组织的错误；而在组织内部，由于灰度值变化不大，只要做简单的线性插值就可以了。基于上述考虑，本文以局部平面来近似组织的表面， 将局部表面方向和梯度信息用于三维灰度图像的内插重建， 充分利用了三维数据的相关信息， 取得了较好的效果。实际的数据实验结果证明， 在保持表面的光滑、连续和清晰的同时，显著减少了内插错误。

Gray scale image interpolation is a preprocessing step for volume rendering in medicine. The rendering quality depends on the result of interpolation directly. Interpolation errors usually show up near the tissue boundary. Since the gray value of pixels on the tissue surface does not change very much, if interpolation is done on the surface of the same tissue, wrong kinds of tissue could not be generated, that is, wrong new tissue resulted from common linear interpolotion can be avoided. While inside the same kind of tissue, because of the small variation of the value of the pixels, linear interpolation will do. Based on the above mentioned consideration local plane is used to approximate the tissue, and local surface directions and gradient information are used for the interpolation of 3 D gray scale images. Since the correlation of 3 dimensional data is used, better results were obtained in respect of smoothness, sharpness, and connectivity and interpolation error is reduced dramatically. Experimental results from real volume data are given.