基于特征模糊推理的形态学颗粒分割算法

Morphology granule segmentation algorithm based on fuzzy reasoning of image features

针对粘连或重叠颗粒图像的分割问题,提出了一种基于特征模糊推理的局部形态学重构参数计算方法,对传统的距离变换结合分水岭的算法进行了改进。在传统距离变换结合分水岭方法的基础上,将颗粒图像划分成若干连通区域,每个连通区域单独处理,使用形态学局部重构的方法抑制分水岭的过分割现象。通过对距离图像连通区域极大值进行统计分析,提取该连通区域的颗粒形态特征。将颗粒形态特征作为模糊输入,重构参数特征作为模糊输出,使用模糊推理方法自适应地计算重构参数,解决了重构参数选取的不确定性问题。最后对重构图像进行分水岭变换得到颗粒分割图像。实验结果表明,该方法对各种粘连状态的颗粒分割效果良好,克服了传统方法的过分割与参数自适应选择的问题。

For the segmentation problem of connective or overlapping granule image,a kind of local morphology reconstruction parameter calculation method was proposed based on fuzzy reasoning of image features which improved the traditional algorithm of watershed combining distance transformation.Granule image was divided into several connected regions based on the traditional algorithm of watershed combining distance transformation,and every connected region was processed separately.Then morphology local reconstruction was used to solve the over segmentation problem.Granule shape features of connected region were extracted through statistically analyzing maximum points of connected region in distance image.Granule shape features were regarded as fuzzy inputs and reconstruction parameter feature as fuzzy output.Morphology reconstruction parameter was adaptively calculated by using fuzzy reasoning which resolved the uncertain problem of reconstruction parameter selection.Finally,watershed transform was carried out on reconstruction image to obtain granule segmentation image.The experimental results show that the improved method can accurately segment various overlapping granules;moreover,conquer the over segmentation problem of traditional method and self-adaptive parameter choice problem.