基于萤火虫算法的三维Renyi熵眼底图像血管分割

对于眼底血管网络分割精度低的问题,提出了基于萤火虫算法的三维最大Renyi熵眼底血管分割方法。该方法先提取出眼底G通道图像;然后用多尺度线性滤波器对眼底血管增强;接着引入萤火虫算法,将基于三维共生矩阵的最大熵求解问题转化为寻找最亮萤火虫的问题;最后,将最亮萤火虫所处的三维空间位置作为Renyi熵函数的阈值对眼底图像分割。实验结果表明,方法的真阳性率和ROC曲线下方区域面积都有所提高,能准确分割出眼底血管。

Three dimensional Renyi entropy segmentation of fundus images based on firefly algorithm

For the fundus vascular network segmentation of low precision, Three-dimensional maximum Renyi entropy fundus segmentation method based on firefly algorithm is proposed. The method firstly extracts the G channel images of the fundus, and then uses multi-scale linear filter to enhance the fundus blood vessels. Nextly, a firefly algorithm is introduced to convert the maximum entropy solution of the Three-dimensional co-occurrence matrix to the problem of finding the brightest firefly. Finally the three-dimensional position of the brightest firefly is used as the threshold of the Renyi entropy function to segment the fundus image. The experimental results show that the true positive rate of this method and the and the area under the ROC curve are improved, The blood vessels of fundus image can be accurately segmented.