基于遗传算法的阈值分割桥梁裂缝检测算法研究

Research on Threshold Segmentation Algorithm of Bridge Crack Detection Based on Genetic Algorithm

针对以超声波原理为基础的非数字化桥梁裂缝检测技术受限于人力与桥梁位置条件的问题,提出一种基于遗传算法的阈值分割桥梁裂缝检测方法。为了改善检测图片质量且兼顾运算效率,在非局部均值滤波算法的基础上提出一种基于积分图像的快速非局部均值滤波算法。结果表明,与传统的中值和均值滤波算法相比,非局部均值滤波算法和基于积分图的非局部均值滤波算法具有更好的降噪效果,但在运行效率上基于积分图的非局部均值滤波算法约是经典非局部均值滤波算法的30倍,能够满足对图像的实时处理。基于此,提出一种基于遗传算法的阈值分割算法对降噪后图像进行分割,可以实现对桥梁裂缝的快速高精度检测。结果表明该方法不仅测量精准、运行效率高,而且操作方便,具有一定的实用价值。

Aiming at the problem that the non digital bridge crack detection technology based on ultrasonic principle is limited by man‐power and bridge position conditions,a threshold segmentation method for bridge crack detection based on genetic algorithm is proposed.In order to improve the image quality and transmission efficiency,a fast non local mean filtering algorithm is proposed based on the non local mean(NLM)filtering algorithm,namely integral non-local means algorithm(INLM).The results show that NLM and INLM algorithm have higher noise reduction effect than the traditional median and mean filtering algorithms,but in terms of operation efficiency,INLM is about 30 times of the classical non local mean filtering algorithm,which can meet the real-time image processing.Based on this,a threshold seg‐mentation algorithm based on genetic algorithm is proposed to segment the image after noise reduction,which can realize the rapid and high precision detection of bridge cracks.The results show that the proposed method is not only accurate in measurement,high in operation effi‐ciency,but also easy to operate,with certain practical value.