属性形态运算及其在心脏图像分割中的应用

The Attribute Morphological Operations and Their Application in Heart Image Segmentation

图像分割在心室容积计算中起着关键作用,文中提出了基于属性形态运算的分割方法。属性形态开闭运算类似于重建开闭,因为它们都是连通成分的变换。然而,属性形态开闭运算更具普遍性,它们描述更一般化的形状特征并且没有形状改变,它们只测试连通成分是否满足某种递增准则,满足的保留,否则移除。将属性形态运算与梯度算子结合,就构成一种强抗噪性的分割方法。将这些运算应用到有噪声的心脏图像的分割,不仅将噪声去除,而且图像的边缘保留完好,取得满意的效果。

Image segmentation plays an important role in computing the volume of heart. The segmenting method based on attribute morphological operation is presented. The attribute opening and closing operations based on connected components are similar to opening and closing by reconstruction since they are connected component transformations. However, the attribute opening and closing operations based on connected components are more general because they can describe generalized shape attributes and their shapes are preserved. They just test whether a connected component satisfies some increasing criterion. If a connected component satisfies the criterion, the connected component remains. Otherwise it can be removed. The good results can be gained by applying these operations in image segmentation because not only the noise is canceled but also the edges stay well.