一种改进非极大值抑制的Canny边缘检测算法

An Canny Edge Detection Algorithm Base on Improved Non-Maximum Suppression

在传统Canny算法的非极大值抑制过程中,由于对比点的选择存在随意性而导致边缘检测不准确。针对此问题,提出了一种改进非极大值抑制过程的Canny边缘检测算法,利用目标像素点梯度方向周围4个像素点和邻域相关系数,在梯度方向上进行插值,代替传统算法中直接以梯度方向周围的邻近像素点作为对比点,实现新的非极大值抑制过程。通过实验证明:改进的边缘检测算法在边缘检测及虚假边缘抑制的性能方面有所提升。

In the non-maximum suppression process of Canny algorithm, the randomness in selecting contrastive points inaccuracy in edge detecting. To solve the problem, this paper proposes improved algorithm based on classical Canny algorithm. The improved algorithm interpolate points based on both four pixels around and the related parameter to replace the pixels near gradient direction as contrastive points, then the new non-maximum suppression process. The results of related experiments show that the improved algorithm has better performance on both edge detection and false edge suppression than the classical Canny algorithm.