基于面积稳定性与形状属性的杆状区域分割*

针对不同光学环境下模糊图像和复杂形状交叠图像中目标区域的分割问题，研究了基于属性形态学分析的杆状区域的滤波和提取方法。应用连通域的面积稳定性和形状属性特征对模糊边界的杆状目标进行滤波，连通域变换中的面积稳定性有效控制粘连细胞等大区域的形成，而形状属性以个体模板目标有效规约连通域的生长和修剪。实验表明：组合属性形态学方法有效地排除了非杆状物等背景干扰，有效分割出的目标像素达90%左右，识别出的目标区域达96%。该方法有效提高了用连通域表示的特定形状区域的分割精度和鲁棒性。

Segmentation of elongated objects based on area stability and shape profiles

For the issue that segmentation of target regions in complex images and blurred images in different optical environments, this paper researched filtering and extracting methods of elongated objects based on attribute morphology analysis. It filtered elongated objects with indistinct borders using area stability and shape attribute. It controls efficiently overlapped cell and bigger connected component forming by the area stability attribute and restrains connected component growing and components max-tree node pruning procedure by shape attribute. It has implemented a accurate segmentation to elongated objects .The experiment show that the proposed method has clear up non-elongated object and other background noises, and the obtained target pixel is about 90%, the target area is up to 96%. The proposed method promotes the segmentation accuracy rate and algorithm robustness.