一种基于梯度方向一致性的边缘检测新算法an-电子与信息学报.pdfVIP

第38 卷第7 期 电 子 与 信 息 学 报 Vol.38No.7 2016 年7 月 Journal of Electronics Information Technology . Jul. 2016 一种基于梯度方向一致性的边缘检测新算法 李 正* 张 海 (北京航空航天大学飞行器控制一体化技术重点实验室 北京 100191) 摘 要：为达到同时提取图像的主要边缘和微弱边缘并有效抑制噪声的目的，该文利用真实图像边缘两侧的灰度渐 变性，以及边缘点周围灰度梯度的方向一致性好而非边缘点周围灰度梯度的方向一致性差的特点构造了梯度方向和 (SGD)指标；并根据该指标提出一种阈值自适应的边缘检测算法。实验表明：梯度方向和在有效提取边缘点的同时 能较好地抑制高强噪声；该指标对光照和对比度变化有较强的鲁棒性；将其用于阈值的自适应调节，得到的基于梯 度方向一致性的边缘检测方法能较好地解决兼顾弱边缘检测的同时而不引入噪声干扰的问题。 关键词：边缘检测；梯度方向；一致性；抗噪；自适应阈值 中图分类号：TN911.73 文献标识码：A 文章编号：1009-5896(2016)07-1674-08 DOI: 10.11999/JEIT151130 An adaptive Edge Detection Approach Based on the Gradient Direction Consistency LI Zheng ZHANG Hai (Science and Technology on Aircraft Control Laboratory, Beihang University, Beijing 100191, China) Abstract: A novel mathematical index about edge detection is constructed to indicate both conspicuous edges and inconspicuous edges in a gray-level image. The index called Sum of Gradient Direction (SGD) is derived from the basic idea that the gradient directions of the points surrounding the real edge point have good consistency while the gradient directions of those surrounding the noise point have poor consistency. According to the SGD index a new adaptive thresholding method to detect edges is proposed. A great quantity of experiments show that: the SGD index has the ability to distinguish both conspicuous edge points and inconspicuous edge points from the noisy points; the proposed novel edge detector utilizing the SGD to regulate the gradient threshold has the ability of detecting weak edges and suppressing noisy points at the same time. Key words: Edge detetion; Grident direction; Cons