改进的基于渗滤模型的混凝土裂纹检测方法

IMPROVED CONCRETE CRACK DETECTION METHOD BASED ON PERCOLATION MODEL

针对传统的基于渗滤模型的混凝土裂纹检测方法设置加速度参数的随机性,而该参数对裂纹检测的准确度有较大影响的问题,提出一种改进方法。该方法利用背景区域和裂纹区域的差异性,结合原始渗滤窗口和最大渗滤窗口的大小自动计算出合适的加速度参数。实验结果表明,该方法能够减少背景像素点和裂纹像素点的误判,避免了原方法中设置加速度参数的盲目性,在裂纹检测的准确度方面有很大提高。

In this paper we propose an improved method for the problems of traditional percolation model-based concrete crack detection method. In original method,the acceleration parameter setting is stochastic,and the parameter has a greater impact on the accuracy of crack detection. The new method uses the differences between the background area and the crack region,combines the sizes of the initial percolation window and the maximum percolation window to automatically calculate the appropriate acceleration parameter. Experimental results show that this improved method can reduce the misjudgement on background pixel points and crack pixel points,avoids the randomness of the acceleration parameter setting in original method,and significantly enhances the accuracy of the concrete crack detection.