基于消除刻槽的水泥路面新生裂缝识别方法

Groove removal based Newborn Crack Reconigition Method of Cement Concrete Pavement

摘要:

利用二维傅里叶变换(FFT)消除刻槽，并通过同态滤波增强裂缝对比度；采用自适应方向的LoG算法在空域中增强裂缝；使用Otsu法，对全局图像进行二值化处理，并利用方格法提取裂缝片段；在此基础上，使用曲线累加变向法计算目标光滑程度，实现“沥青拉丝”的剔除.试验表明，本文方法具有良好的识别效果.

Abstract:

Two dimensional Fourier transform(FFT) was adopted to remove groove, and a homomorphic filter algorithm was employed to enhance the crack contrast in frequency domain. An adaptive direction LoG algorithm was employed to highlight the crack. The Otsu method was resorted to processing the global image into binary images and the grid method was adopted to extract crack fragment. Finally, the degree of target smooth was calculated with the curve cumulative turning method, and the removal of the “asphalt drawing” was realized. Experiments prove this method to be of good recognition effect.