提取图像边缘过程中抑制噪声的随机启发搜索算法

常用的边缘提取方法对噪声的抑制能力不强,导致了大量不连续边缘的产生。中提出了一个随机启发式搜索算法,它能够有效地抑制噪声,从而连接到连续的边缘,同时,介绍了随机启发式搜索算法,讨论了起始点的选择准则和扩展点的选择方法,最后介绍了终止点选择的方法。一个起始点,一个终止点和许多扩展点就形成了一个候选边缘,通过候选边缘的随机积累进行自增强,达到抑制噪声,提取清晰边缘的目的,中使用加入高斯噪声的“Lena”图像作为实验图像,实验结果初步证明了该算法能够提取到连续,清晰的边缘。

A Stochastic Heuristic Search Algorithm for Suppressing Noise in Edge Extraction

Usual methods for extracting edges are not particularly good at suppressing the noise; as a result discontinuities in extracted edges often occur. We propose a stochastic heuristic search algorithm that can suppress the noise, thus linking edge points effectively to form a continuous edge. Section 1 explains what is stochastic heuristic search. Subsection 1.1 discusses the selection of initial point for starting search. Subsection 1.2 discusses the selection of extension point for further search. Subsection 1.3 discusses the end point of search. One starting point, one end point, and very many extension points form a candidate edge trajectory. Many randomly accumulated candidate edges reinforce each other mutually and form an extracted edge that suppresses noise very efficiently and thus is very distinct. Our experimental image is "Lena" with Gaussian noise added by us (Fig.2). The edges extracted from this experimental image, shown in Fig.3, demonstrate preliminarily that our stochastic heuristic search algorithm can extract continuous and distinct edges.