【作者】 曹永锋； 孙洪； 徐新；

【Author】 CAO Yong-Feng, SUN Hong, XU Xin School of Electronic Information, Wuhan University, Wuhan, 430079, China,

【机构】 武汉大学电子信息学院通信系；

【摘要】 在图像处理和模式识别的应用中常需要一幅图像在不同约束下的多个分割结果。以弧段动力学作为区域轮廓的显著性评价,提出了一种多等级区域轮廓提取的高效方法。该方法融合流域分割,计算盆地动力学(Basin dynamic),和计算弧段动力学(Arc dynamic)三个过程于一次溢流过程之中:一次处理得到所有区域轮廓的显著性,通过简单阈值操作可得到不同显著性级别的单像素闭合区域轮廓。更多还原

【Abstract】 In image processing and pattern recognizing, different segmentations of one image under different restriction are often needed. Using arc dynamic as a measurement of contours’ saliency, one hierarchical contours detection method of high efficiency was proposed. The method merge watershed segmentation, calculating basin dynamic and calculation arc dynamics in only one flooding process, and get saliency value of all possible field contours. With different thresholds, one-pixel-width close contours of different saliency level can be easily detected.更多还原