一种改进的基于距离变换的图像目标轮廓插值方法

An Improved Distance Transform-Based Contour Interpolation Methodfor Image Object

本文提出了一种改进的基于欧几里德距离变换的图像目标轮廓插值方法。首先由获取的层间间距相对较大的序列断层图像,经由对图像中目标的轮廓跟踪和对图像中目标的区域填充获得图像目标的轮廓图和区域图,采用欧几里德距离变换求取该序列断层图像目标轮廓的距离图,运用逆距离加权平均线性插值方法求得相邻图像中目标的距离图之间的轮廓图像。运用本文方法得到的中间插值轮廓过渡平滑,并且能够很好地解决极端凹轮廓之间的中间插值轮廓问题,本文的方法具有较高实际应用意义。

This thesis proposes an improved contour interpolation method based on the Euclidean distance transform for image object. Firstly, we get the contour images and region images by tracking and filling the object regions in the images which are acquired with relatively sparse spacing, then we get the distance images of the serial contour images using the Euclidean distance transform, as a result, we get the corresponding intermediate interpolation contour images using the method of linear interpolation. The transition between slices which are attained by applying the proposed method is more smooth. Moreover, the method perfectly solves the interpolation between extremely concave contours. The proposed method has highly practical significance.