摘要
基于桥梁图像的病害分析是对桥梁日常巡检的主要手段,其中桥梁图像的病害以各种形式的裂缝为主。为提高对桥梁表面裂缝情况的检测精度,提出一种基于k-means和概率松弛算法的图像分割方法。首先,利用k-means算法能够快速和有效地处理大量数据的特点,根据灰度值将含有裂缝的数字图像粗分成三类;然后,利用概率松弛算法并行计算和含有裂缝的数字图像像素点之间的空间结构特点,对含有疑似裂缝信息和背景的类做细提取,更精确的判断像素点属于裂缝还是背景;最后,通过对多种桥梁图像的处理进行算法性能的检测,实验结果表明,相对于对比算法,算法在对噪声点的抑制、裂缝边缘的细化上均有较好的效果,并对光照具有一定的鲁棒性。
Analysis of defects based on bridge images is the main bridge routine inspection method, the disease of bridge images is mainly in various forms of cracks. In order to improve the detection accuracy of the cracks on the surface of the bridge, an image segmentation method based on k-means and probability relaxation algorithm is proposed. Firstly, split the digital images with bridge cracks into three categories according to the grayscale value. Then, use parallel computing and digital images considering the spatial structure of the probabilistic relaxation algorithm to further optimize the classification of the suspected fracture and background. Finally, the performance is tested by processing a variety of bridge images. The experimental results show that, compared with other images segmentation algorithms, the proposed algorithm is better in robustness to noise, illumination, and thinning of crack edges.
出处
《控制工程》
CSCD
北大核心
2017年第S1期146-150,共5页
Control Engineering of China
基金
国家自然科学青年基金(61502423)
浙江省科技厅科研院专项(2016F50047)资助
