【摘要】 针对视网膜图像中因复杂的脉管结构、光照不均等因素造成由一般的阈值处理方法难以准确分割病灶区域的问题,提出了一种基于修改形态学区域深度信息的视网膜图像分割方法。该方法基于形态学中的区域极值概念。从区域极值的特征分析入手来修改极值区域,最后进行扩展的H-极小值变换实现图像的区域分割。实验结果表明,该方法能同时解决过分割和欠分割问题,在准确率和检全率上都得到了满意的病灶分割效果。更多还原

【Abstract】 General threshold approach for retinal image segmentation has the difficulty to achieve accurate focus area due to complex vascular structures and uneven illumination,etc. To obtain precise segmentation result,a method based on morphological region depth information is used on the retinal image segmentation. The method is based on the morphology of the regional extreme concept. Starting from the characteristic analysis of the regional extremes it modifies extreme region. Finally extended Hminimum transform realizes the regional segmentation of images. The experimental results show that the method can solve the problem of over-segmentation and owe segmentation,on the accuracy and recall rate of segmentation results are satisfied the lesions.更多还原