【摘要】 对传统的微分边缘检测算法的缺陷进行分析和评价,并针对微分边缘检测算法的缺陷,提出了一种改进的边缘检测算法。新算法摈弃了传统边缘检测算法中将每个像素单独进行考虑的模式,而是将待检测像素及与其相邻的像素作为一个整体来考虑,通过将待检测像素为中心的某个领域进行整体计算,得出能够反映这个领域性质的某些参数,并以此作为边缘检测的判据,进行检测。此算法充分考虑到了中心像素与周围像素之间的关系,从而增加了梯度运算的可靠性。实验表明:新算法能够检测到更细的边缘,并且在椒盐噪声的情况下要优于常用的微分边缘检测算法。更多还原

【Abstract】 The defects of conventional gradient edge detection are analyzed and evaluated.In order to avoid the defects of gradient operators,this paper proposes a new edge detection method.The new algorithm alters the traditional method of considering one pixel separately by considering the neighborhood of one image pixel as a whole part.Some special parameters of the area can be presented through calculating the pixel and its neighborhood.The edge detection is based on those parameters.The new algorithm thoroughly considers the neighborhood of one image pixel and improves the stability of the gradient results by introducing the value of the neighborhood to the gradient.The edge detection results of test image show that the new method can detect thinner edges and is more reliable under salt and pepper noisy condition.更多还原