文章摘要
引用本文:林丽群,王卫星.改进的分数阶微分及图论的粘连血细胞图像分割[J].福州大学学报(自然科学版),2017,45(6):794~800
改进的分数阶微分及图论的粘连血细胞图像分割
Touching blood cell image segmentation based on improved fractional order differentiation and graph theory
  
DOI:10.7631/issn.1000-2243.2017.06.0794
中文关键词: 血细胞图像分割  分水岭算法  最小生成树(MST)  分数阶微分
英文关键词: blood cell image segmentation  watershed algorithm  minimum spanning tree(MST)  fractional differentiation
基金项目:
作者单位
林丽群 福州大学物理与信息工程学院福建 福州 350116 
王卫星 福州大学物理与信息工程学院福建 福州 350116 
摘要点击次数: 116
全文下载次数: 96
中文摘要:
      针对血细胞图像模糊及对比度不高的现象,提出一种改进的分数阶微分的图像预处理方法. 即将形态学去噪和改进的类圆形掩膜算子的分数阶微分增强结合起来,在滤除血细胞图像的染色污染和颗粒噪声的同时较好地保留了细胞边缘细节. 针对分水岭算法存在的过分割和最小生成树算法存在的效率较低问题,采用分水岭算法和最小生成树算法相结合的图像分割算法. 首先用分水岭算法初分割分数阶微分增强的细胞图像,接着算法选取过分割区域映射为节点,最后基于改进的最小生成树算法再分割细胞图像. 实验表明,该算法能有效缓解分水岭算法的过分割,并且有效减少了最小生成树算法中节点的数目,提高算法效率.
英文摘要:
      For blurring and low contrast of the blood cell images,an image preprocessing method based on improved fractional differential was put forward. This method was combined with morphological denoising processing and improved round mask operator of fractional differential enhancement,filtering the staining contamination and grain noise of the blood cell images,and retaining the edge details of the cells. Also an improved cell image segmentation algorithm based on watershed algorithm and minimum spanning tree (MST) algorithm was proposed,which solved the problem of low efficiency and over-segmentation. The pre-processed cell images were split by watershed algorithm. Then,the over-segmentation areas were mapped to nodes. At forwards,the cell images were segmented by the improved MST algorithm. The new algorithm can alleviate the over-segmentation problem of watershed algorithm,and effectively reduce the number of nodes in MST algorithm. Experimental results indicate that the studied algorithm is ideal and effective.
查看全文   查看/发表评论  下载PDF阅读器
关闭