基于形状相似性和曲线化简的统计形状模型建立

Building statistical shape models based on shape comparability and curve simplification

针对活动形状模型(ASM)中统计模型建立存在的标记点的自动选取和点的对应两大问题,提出了一个完整的解决方案:首先通过矩阵的相似变换对齐训练集中各样本,然后根据样本特征,选择基准样本,再利用曲线化简方法在基准样本上自动提取特征点,最后利用训练样本间形状的相似性,由基准样本上的特征点向其它样本投影得到所有样本上的标记点,实现了标记点的自动选取,同时保证了样本间标记点的一一对应。为了验证该方法的精确性和有效性,分别建立股骨和髋臼骨的点分布模型,实验结果表明,模型形状变化表现力强、计算效率高。分别与传统的手工标记法和优化法比较,在具有相同形状表现力的情况下,本文方法建模效率更高,时间开销更小。

In the active shape model（ASM） ,the key to build the point distribution model（PDM） is automatic. We provide a solution for this problem which consists of： （1） aligning the training set,（2） selecting the basic sample in the training set, （3） a process of automatic landmark selection for the basic sample, and （4） a process of propagating landmarks on each training sample for defining landmarks in them. The method is evaluated on hip joint data sets and the objects of interest are the femur and acetabulum bone. The results indicate that,better compactness of the ASM can be achieved. The method is also evaluated on other training sets of shapes,and the results show that it constructs better models than alternative approaches.