基于边界信息的医学图像三维插值

目的Cubic卷积插值是医学图像三维插值的常用方法,针对其插值处的结果边界模糊和精度不高的缺陷,建立一种精确度较高的插值方法。方法首先通过模糊对比度增强精确定位图像边界,再运用形态学运算确定出新插值图像边界,对于新插值图像边界点采用最佳匹配对应点插值;对于非边界点采用一种新的Cubic卷积插值方法确定其灰度值。结果本文方法的均方差、不符合像素点数和最大误差均小于传统插值方法。结论本文提出的方法具有较高的精确性。

Three-dimensional interpolation of medical images based on boundary information

Objective Cubic convolution interpolation method is usually used for the three dimensional interpolation of medical images, but this approach can bring on a fuzzy boundary of the interpolated image and low accuracy. Aiming at the deficiencies, a method is presented with higher precision. Methods Firstly, the contrast of images was enhanced to locate the boundary well and truly, and then determined the boundary of the interpolated image using the mathematical morphology operators. Finally, value of the boundary pixels was interpolated by the best matching corresponding points. At the same time, to the pixels not on the boundary of the interpolated image, the grey value was obtained by a novel Cubic convolution interpolation which was developed in this paper. Results Mean-squared difference, number of sites of disagreement and largest difference of the developed method were less than the conventional methods. Conclusion The approach proposed hither performs better than the conventional Cubic convolution interpolation and linear interpolation.