摘要
眼底出血区检测在疾病诊断中具有重要意义,计算机处理眼底图像可以减少医生的重复劳动。但由于受到眼底图像质量、检测算法,以及出血区的多样性和复杂性等因素的影响,目前的检测方法存在检测种类粗糙和检出率低的问题。文中提出一种新的眼底病灶检测分割算法,算法包括两次分割和两次标记提取。第一次分割是对图像进行简单的最大类间方差分割,主要是去除大部分的背景,提取分割得到的图像粗标记;第二次分割,主要是对形态学处理后的图像进行连通标记,进行类聚分割,以获得更细致的病变渗血区域。实验结果表明该算法是有效的。
Subhyaloid hemorrhage detection in the diagnosis of diseases is important, fundus image processing can reduce the computer doctor of duplication. But due to retinal image quality, detection algorithms, as well as the diversity and complexity of the bleed area and other factors, the cunnt detection methods have kinds of rough and low rate of detection. It presents a new segmentation of fundus oculi lesion detection algorithm, including two segmentation and two marked extraction. Division for the first time is the simple segmentation of maximum cluster variance for image, mainly removing most of the background, extracting image segmentation tags. The second division, mainly to connect tag, the processed image is morphology for segmenting type,in order to obtain more detailed bleeding lesion area. Ex- perimental results show the effectiveness of the algorithm.
出处
《计算机技术与发展》
2014年第4期207-209,213,共4页
Computer Technology and Development
基金
广东省大学生创新训练项目(1057312025)
广东药学院大学生创新训练项目(2012-20)
关键词
眼底病灶检测
图像增强
阈值分割
形态学处理
类聚分割
fundus oculi lesion detection
image enhancement
threshold segmentation
morphological processing
cluster segmentation
