中文摘要：

为了检测桥梁混泥土表面的裂缝,提出了基于图像的种子游走二值化桥梁裂缝检测算法.首先,随机产生一些种子点,使每个种子点在给定的规则下进行游走；接着,统计这些种子点的位置,将原始图像转化为二值化图像；然后,提取二值化图像的连通分量,构造连通分量的特性指标,消除噪声像素；最后,分析连通分量间的位置关系并进行拼接,提取出桥梁表面图像中的裂缝.试验结果表明,裂缝检测算法具有稳定性高、对光照不均匀以及背景噪声的影响不敏感、检测效果好等特点.

英文摘要：

In order to detect the cracks on the concrete surface of bridge,the image-based seed walking binarization bridge crack detection algorithm is proposed.Firstly,some of the seed points are produced in random,and each seed is walking in accordance with given rule,the positions of these seed points are added up,and the original image is converted into binary image.Then,the connected component of binary image is extracted to build the characteristic index of the connected component and eliminate noise pixel.Finally,the positional relationship among connected components is analyzed and spliced; thus the cracks in image of bridge surface are extracted.The experimental results indicate that this crack detection algorithm features high stability and good detection effect,as well as it is insensitive to the uneven illumination and background noise.