基于WFSOA的2D-Otsu钢轨缺陷图像分割方法

针对二维最大类间方差阈值法(2D-Otsu)抗噪性较弱、计算时间较长的问题,文中提出了一种基于随机权重及异步价值因子取值的人群搜索算法,并将其应用于2D-Otsu中对钢轨缺陷图像进行分割。该算法采用随机权重加快收敛速度,采用异步价值因子提高搜索能力,有利于全局收敛到最优值。根据测试函数分析,WFSOA算法能够快速收敛,寻优值结果精度高,收敛时间短,算法稳定性好。在钢轨缺陷图像分割中,将2D-Otsu的迹函数作为WFSOA的目标函数,实验结果表明图像检测实时性高,对表面灰度不匀或生锈的钢轨缺陷分割结果清晰,有效降低了钢轨缺陷误检率和漏检率,在计算时间上仅占2D-Otsu算法的2%,可满足实际工程对实时性的需求。

2D-Otsu Rail Defect Image Segmentation Method Based on WFSOA

Aiming at the problem that the two-dimensional maximum inter-class variance threshold method(2D-Otsu)had weak anti-noise and long calculation time,a seeker optimization algorithm based on random weight and asynchronous value factor is proposed.The algorithm is applied to the image segmentation of rail defects in 2D-Otsu.The random weight is used to speed up the convergence speed of the algorithm,and the asynchronous value factor is used to improve the algorithm's search ability,which is conducive to global convergence to the optimal value.Through the test function analysis,the WFSOA algorithm can converge quickly,the precision value of the optimization value is high,the convergence time is small,and the algorithm has good stability.In the image segmentation of rail defects,the 2D-Otsu trace function is used as the objective function of WFSOA.The experimental results show that the image detection has high real-time performance,the segmentation result of the rail defects is clear,and the false detection rate and missed detection rate of the rail defects are effectively reduced.Time is only 2%of the 2D-Otsu algorithm,which meets the needs of actual engineering.