改进遗传规划的混凝土路面裂缝检测算法

Concrete surface cracks detecting algorithm based on improved genetic programming

针对在光照复杂环境下裂缝检测困难的问题,提出改进遗传规划的混凝土表面裂缝检测算法,将训练集图像进行光照的归一化处理,消除光照影响,根据裂缝图像特征筛选出训练集,利用改进选择算子的遗传规划算法训练图像处理模型,使用图像处理模型对图像裂缝进行提取并去除块状干扰得到最终检测结果。仿真结果表明,改进选择算子能在保证裂缝检测精度的情况下,加速图像处理模型的训练时间,能够精确、快速、有效地检测出混凝土表面裂缝。

Aiming at the difficulty of crack detection in complex illumination environment, an improved genetic programming crack detection algorithm for concrete surface was proposed. The training set was normalized by illumination, so as to eliminate the illumination effect. The training set was filtered out according to the characteristics of the crack image, and the genetic programming algorithm of the improved selection operator was used to train the image processing model. The image processing mo- del was used to extract the image cracks and remove the block interference to get the final test result. Simulation results show that the improved selection operator can accelerate the training time of the image processing model. The algorithm proposed has strong robustness, and the real concrete surface crack can be detected accurately and effectively.