基于图像连通域的桥梁裂缝提取方法

Bridge Cracks Extraction Method Based on Image Connected Domain

裂缝是水泥混凝土桥梁的一种常见病害形式,基于图像的桥梁裂缝病害检测是实现桥梁数字化无损检测的一项新技术。为此,提出了一种基于图像连通域的桥梁裂缝提取方法,以实现从桥梁图像上准确识别裂缝,并提取其特征参数。针对水泥混凝土桥梁图像采集时,施工接缝、光照、污损、抖动等因素造成的干扰,主要分3个步骤实现裂缝目标的提取:(1)对所采集的经过去模糊处理的桥梁图像进行特征分析;(2)根据像素灰度对比提取疑似裂缝信息;(3)利用连通域度量去除虚假裂缝,保留真实裂缝信息。通过对大量的桥梁图片进行测试,结果表明,该方法能实现桥梁真实裂缝特征的提取,具有较强的鲁棒性和实用性。

Cracking is a normal form of cement concrete bridge defects.Image based bridge crack detection is a new technology to realize digital non-destructive test for bridge defects.A method for bridge cracks extraction based on image connected domain was proposed for recognizing and extracting the characteristic parameters of the cracks on the bridge surface.Three main steps of cracks extraction aimed at resolving the interference of construction joints,sunlight,defacements and vibration were performed：（1） the cracks＇ characteristics of the acquired and denoised bridge images were analyzed;（2）suspicious cracks were extracted by the method based on pixel grey-value comparison;（3） the false cracks were removed by using the computed measurement of connected domain,and the true ones were reserved.Quantities of images were tested,and the results show that the proposed method,which has good robustness and availability,can extract the bridge cracks＇ characteristics actually.