【摘要】 眼底视网膜图像的血管增强与分割在疾病的诊断与预防中具有广泛的研究价值。为了较准确地识别细小血管,方便医生诊断,提出了一种基于主成分分析(PCA)的血管分割算法,先对眼底图像提取特征空间,然后使用PCA对特征空间进行特征提取,最后得到一幅关于眼底特征空间的特征图,为了消除特征图的噪声和突变,得到较清晰的分割图像,首先对此特征图进行局部均值化操作,最后使用阈值进行分割。在DRIVE数据库中的眼底图像上进行实验,并与现有的非监督算法进行对比,实验结果表明,该方法具有较大的准确性。更多还原

【Abstract】 Retinal image enhancement and segmentation in the diagnosis and prevention of disease has a wide range of research value.This paper proposed a blood vessel segmentation algorithm based on principal component analysis. Firstly, the feature space is extracted from the fundus images, then a is obtained by using PCA. The local mean operation is played on the feature map, finally the segmented image is obtained by using threshold. An experiment was carried out in the DRIVE database, and was compared with the existing unsupervised algorithms, experimental results show that the method has great accuracy.更多还原