基于支持向量机的多光谱显微细胞图像分割  

Support Vector Machines for Multi-spectral Microscopic Cell Image Segmentation

作　　者：李美娟[1] 王文伟[1] 杨定楚[1] 王思贤[1] 

机构地区：[1]武汉大学电子信息学院,武汉430079

出　　处：《计算机工程与应用》2006年第8期37-39,43,共4页Computer Engineering and Applications

基　　金：教育部留学回国人员科研启动基金资助项目(编号:212152457)

摘　　要：文章根据多光谱图像数据维数高的特点,以像素各个波段的灰度值为特征,采用支持向量机(SVM)方法为核心来分割多光谱显微细胞图像。为提高计算速度,在亮度和色度分量上利用阈值分割法进行预处理;同时,对SVM分割后的图像,采用基于区域灰度差的生长准则进行后处理,获得了更好的分割效果。最后把支持向量机(SVM)方法与径向基神经网络(RBFNN)方法进行比较,实验结果表明,SVM分割效果优于RBFNN,是一种精度高、速度快的多光谱显微细胞图像分割方法。According to the high-dimension data of multi-spectral microscope cell image,an image segmentation method based on Support Vector Machines（SVM） is presented,which uses gray values of the pixel＇s bands as features. Meanwhile pre-process using the threshold segmentation method on the intensity weight and hue weight is adopted to increase the computational speed and post-process using region growing criterion based on gray value difference is adopted to get a better segmentation result.Experiment shows that the SVM-based segmentation method is suitable for Multi-spectral Microscopic Cell Image.It can segment the Muh-ispectral Microscopic Cell Image more efficiently and precisely than RBFNN-based method.

关 键 词：多光谱图像 细胞图像分割 支持向量机 径向基函数 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]