- 上一篇：云计算服务OpenStack的代数规约描述技术
- 下一篇：asp.net+sqlserver网络新闻采集系统设计
摘要RPE层是基于频域光学相干断层（SDOCT）视网膜图像的最后一层，也是七个视网膜层中最亮的一层。精确分割RPE层对眼部疾病的诊断和研究来说非常重要。然而，手动分割是一个耗时和主观性很强的过程，而且需要专家来完成。本文利用图论的知识，提出了一个概念简单且容易实现的自动分割SDOCT图像RPE层的算法。算法包括图像降噪，平整化处理，搜索区域限制，构造无向图并求权重，用Dijkstra 算法求图的最短路径等几个步骤。其中，平整化处理步骤最耗时，搜索区域限制步骤对分割结果的精确性影响最大。实验结果表明，本论文提出的算法能高效精确的分割RPE层。65993
毕业论文关键词 图像分割； 图论； RPE层； 视网膜； SDOCT图像
毕业设计说明书（论文）外文摘要
Title Segmentation of RPE layer in Spectral Domain Optical Coherence Tomography images
Abstract
RPE layer is the bottom and the brightest layer of seven retinal layers in Spectral Domain Optical Coherence Tomography (SDOCT) image. Accurate segmentation of RPE layer is crucial for the diagnosis and study of ocular diseases. However, manual segmentation is often a time-consuming and sub- jective process, and meanwhile, it needs an expert grader to do this. This paper presents a conceptually simple, yet easy to implement, automatic approach for segmenting RPE layer in SDOCT images using graph theory. This approach includes several steps, such as Image Denoising, Image Flatten, Search Space Limitation, Create Graph And Calculate Graph Weight, Get The Shortest Path In Graph and so on. Among all these steps, Image Flatten is the most time-consuming step and Search Space Limitation has the great- est impact on the segmentation results. Experimental results demonstrate that the algorithm proposed in this paper is able to segment RPE layer efficiently and accurately.
Keywords: image segmentation; graph theory; RPE layer; retinal; SDOCT
images
目录
1 绪论 1
2 自动分割视网膜SDOCT图像RPE层的算法简介 3
2.1 图形表示和权重计算 3
2.2 层端点初始化 5
2.3 搜索区域限制 6
2.4 最短路径求解 6
3 自动分割视网膜SDOCT图像RPE层的算法详细实现 6
3.1 图像降噪 9
3.2 图像平整化与RPE层曲率消除 10
3.3 搜索区域限制 12
3.4 权重求解 18
3.5 代表RPE层上下边界的单源最短路径的求解 19
4 实验结果分析 23
4.1 自动分割与手动分割的对比 23
4.1.1 自动分割与手动分割的定性对比 23
4.1.2 自动分割与手动分割的定量对比 26
4.2 算法的执行效率分析 30
结论 30
致谢 31
参考文献 32