针对眼底图像中血管与背景间对比度低以及血管自身结构复杂等因素对视网膜血管分割所带来的问题, 本文提出了一种具有自适应连接值的脉冲耦合神经网络(PCNN)与高斯匹配滤波器相结合的视网膜血管分割方法。首先, 利用对比度受限制的自适应直方图均衡化(CLAHE)技术与二维高斯匹配滤波器对血管区域的对比度进行有效增强。然后, 利用经验阈值选择出一定的血管区域作为初始种子区域。接着, 将带有快速连接机制的PCNN与种子区域增长思想相结合, 通过自适应动态设置PCNN中的连接强度系数和停止条件, 实现眼底图像中血管区域的自动生长。整个算法在DRIVE视网膜图像库中进行了测试, 实验结果表明, 相比于不使用动态连接强度系数与停止条件的方法, 所提出算法的灵敏度从49.79%提高至70.39%, 准确度从95%提高到95.39%。证明了该算法具有较好的分割精确度和应用价值。

According to the several problems on retinal vessel segmentation caused by many factors such as low contrast of vessels and background in fundus image and the complicated structure of blood vessel, a retinal vessel segmentation method combining the Pulsed Coupled Neural Network (PCNN) which has self-adaptive linking value, with the Gaussian matched filter was proposed in the paper. Firstly, the contrast of the retinal vessel area was effectively enhanced by application of Contrast Limited Adaptive Histogram Equalization (CLAHE) and two-dimensional Gaussian matched filter. Then, a certain blood vessel area was selected as primary seed area by experience threshold. On the basis of above, the PCNN contained fast linking system was combined with seed area growth thought to set the linking strength coefficient and stopping conditions of PCNN according to the self-adaptive dynamics, result in the automatic growth of vessel area in fundus image was realized. The whole algorithm was tested in DRIVE image database, the results show that, as compared with the method not using the dynamic linking strength coefficient and stopping conditions, the proposed algorithm has had its sensitivity increased from 49.79% to 70.39% and the accuracy has been increased from 95% to 95.39%, which proves that such algorithm has better segmentation accuracy and application values.