中文摘要：

在牙种植技术中,牙齿特征点的有效提取对后续的三维配准和重建具有重要的意义,现有方法的计算效率比较低;论文改进了离散曲线演化算法,采用曲线特征因子量描述牙齿CT各断层图像边缘曲线的复杂性,并根据曲线特征因子量自适应确定不同层间图像边缘曲线特征点提取的数目,以降低数据存储的冗余量,提高特征点的提取效率;用改进的离散曲线演化算法对牙齿不同层的临床CT图像提取特征点,并将实验结果与现有离散曲线演化算法的结果进行比较。结果表明,改进方法在提取每层CT图像特征点所需时间约为原算法的50%,同时提取的特征点数约为原算法的80%。将改进方法提取的特征点按不同比率进行三次样条曲线插值并进行后期重建,其重建效果能很好地反映牙齿的真实结构。因此,改进方法的计算效率远高于离散曲线演化算法,在牙齿种植领域中具有临床应用前景。

英文摘要：

Extraction of tooth feature points effectively is of significance for the subsequent three-dimensional registration and reconstruction.However,the efficiency of current calculation methods is low.This work improved discrete curve evolution（DCE） algorithm and used characteristic factor of curve（CFOC） to describe the complexity of contour on each dental CT slice.With the adaptive CFOC,the feature points number in curves in different slices could be determined,thus the redundancy in the stored data was reduced,improving the extraction efficiency of feature points.The improved DCE algorithm was used for exacting the feature points from different clinical CT slices,comparing experimental results with that of the conventional DCE algorithm.The results showed that the improved method saved nearly half of time that the conventional method required for extracting feature points from each CT slice,and the feature points number extracted by the improved method is approximately four fifths of that of the conventional method.After cubic spline interpolation with different ratios of the feature points from different slices,the slice images were reconstructed and the reconstruction results reflected the true structure of the teeth.Therefore,the calculation efficiency of the improved method is much higher than that of the conventional DCE algorithm.