图像边缘提取的启发式搜索算法

常规的图像边缘提取方法容易受到噪声的影响,导致了大量不连续边缘的产生。本文试验了一种基于边缘分段自增强的启发式搜索(EdgeDetectionalgorithmbasedonHeuristicSearchED-HS)算法,该算法通过对边缘检测后图像的遍历搜索获得各种可能的边缘轨迹,然后对每条搜索轨迹进行分段自增强,最后根据增强程度的积累决定图像边缘。文中阐述和分析了算法的机理,并以加入了高斯噪声的“Lena图像”作为样本进行了计算验证。结果初步证明了该算法能够在较高的噪声条件下有效的提取到连续、清晰的边缘。

A Novel Edge Detection Algorithm Based on Heuristic Search Methods

The conventional algorithms for edges detection are proved to be more sensitive, and result in the discontinuous contures, while there is additive noise. A new edge detection algorithm that is based on heuristic search method is proposed and partialy proved in this paper, the first step of this algorithm is as usual, the petential edge and sub-edges is detected; then every single sub_edge is searched and reinforced. Finally the exact edge is checked out according to the accumulation of the self-reinforces degree . The mechanism and principle of the algorithm is expressed and analysed,a image "Lena"with additive Gaussian noise is used as test sample, the new algorithm is compared with the conventional ones in the experiment. The result shows that the new algorithm is robust and effective.