基于自组织映射的snake模型.pdfVIP

中国科技论文在线 基于自组织映射的 snake 模型 王平，丁曙光** （合肥工业大学机械与汽车工程学院，合肥 230009） 5 摘要：对于主动轮廓模型（ACM ），本文提出了一种新型的自组织映射（SOM ）为基础的 方法，被称为 BSOM ，即试图结合 SOM 和 Snake-based ACMs （基于蛇形的主动轮廓模型） 的优势，以提取所需的图像中的轮廓。采用特征点，在边缘图的形势下（从一个标准的边缘 检测操作获得），以引导轮廓（如在 SOM-based ACMs 的情况下）随着在局部区域中梯度 和强度的变化，以确保轮廓不会由于特征点故障（弱或破损的边缘）而“泄漏”到的物体的边 10 界。随后运用 BSOM 对灰度级图像进行了验证。测试了 BSOM 对图像上的对象具有凸和凸 边界的处理效果。结果表现出 BSOM 的优越性。最后，对其优越性进行了总结。 关键词：主动轮廓模型（ACMs ）；自组织神经网络（SOM ）；时间自适应自组织映射（TASOM ） 中图分类号：T-012 15 Based on self-organizing map snake model Wang Ping, Ding Shuguang (Hefei University of Technology, HeFei 230009) Abstract: For active contour model (ACM), we propose a new type of self-organizing map (SOM) based approach, called BSOM, namely trying to integrate the SOM and Snake-based ACMs 20 advantage to extract the desired image silhouette. We use the feature point, in the form of the edge map (from a standard edge detection operation to obtain), to guide contour (SOM-based ACMs as in the case) in the local area with the changes in intensity gradient and to ensure contour Since the feature point does not malfunction (weak or broken edges) with a leak into the boundary of the object. And, Snake-based ACMs are different, we do not use an explicit energy function (based on 25 the gradient or intensity) for controlling the motion profile. We deal with BSOM contour extracting a plurality of objects, by dividing a single contour contour as many sub-objects in the image. We use the grayscale image BS