路面裂缝识别中边缘检测间断性的改进

Improvement of Edge Detection Discontinuity in the Pavement Crack Recognition

针对传统边缘检测算子得到图像的间断和不连续的特点,结合基于形态学的图像膨胀算法和图像细化算法对路面裂缝图像的边缘检测进行了改进。先介绍了几种常用的边缘检测算子,并利用各个算子对路面裂缝图像进行了边缘检测,将结果进行了对比,根据对比结果选出了Soble算子为本次实验所用的边缘检测算子,在其基础上改进。最后,根据该类路面裂缝图像的特点,改用"菱形"结构元素代替传统的"方形"结构元素,将间断的路面裂缝图像边缘处理成连续的清晰的边缘,达到了很好的效果。

The image obtained by the traditional edge detection operator has discontinuous and discontinuous. Combination based on morphological image expansion algorithm and image thinning algorithm, the image edge detection of cracks in the road gets improvements. First, this article describes several commonly used edge detection operator. Pavement crack image edge detection, the results were compared. According to the comparison results, I elected soble operator with the experimental edge detection operator, and then made improvements on its basis. Finally, according to the characteristics of such cracks in the road image, I used the ＂diamond＂ structure element instead of the traditional ＂square＂ structural elements of the intermittent pavement crack image edge processing continuous and sharp edges, to achieve good results.