神经切片染色图像边缘的二次提取方法研究  

Research of Twice Edge Extraction for Section Staining Image of Nerve

作　　者：唐铖[1] 唐平[1] 张毅[2] 陈松龄[2] 胡智魁[1] 

机构地区：[1]广东工业大学自动化学院,广东广州510006 [2]中山大学附属第一医院,广东广州510080

出　　处：《计算机技术与发展》2011年第8期89-92,共4页Computer Technology and Development

基　　金：广东省自然科学基金项目(9151008901000006);广东省医学科研基金项目(A2009173)

摘　　要：针对神经切片染色图像的边缘模糊不连续、色泽不均匀、区域不规则的特点,提出了一种效果良好的神经切片染色图像边缘提取方法。通过改进的均值漂移滤波改善图像质量,获得第一次的连续边缘;考虑到切片图像的复杂性,对提取得到的连续边缘运用区域填充及数学形态学的方法处理后进行第二次边缘提取,从而提取到最终的较为准确、光滑的边缘;最后在人体周围神经组织切片图像上的实验证明,该方法具有提取效果好、抗噪声能力强等特点。Aiming at the characteristics of section staining image of nerve such as： edges fuzzy discontinuous,color and luster is uneven,regional irregular characteristics,a good effect edge extraction method is proposed.Image quality is improved by the improved mean shift filtering.By using filtered edge extraction of Canny operators,preliminary continuous edge can be got.Thus extract the final relatively accurate,smooth edge by extracting image obtained using the preliminary continuous edge area filling and mathematical morphology method processing,and again for Canny edge extraction considering the complexity of slice.Finally the result of the experiment in section staining image of human peripheral nerve shows that this method has good extraction effect and anti-noise ability.

关 键 词：边缘提取 改进的均值漂移 神经染色切片图像 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]