一种模板匹配与高适应性的裂缝骨架提取算法

在对裂缝图像骨架进行提取时，已有的算法通常存在细化后骨架主体信息缺失、毛刺去除效果随图像规模增大而快速下降等问题。针对上述问题，该算法提出一种模板匹配与高适应性的裂缝骨架提取算法。首先，结合模板匹配对Rosenfeld细化算法进行改进，以保留骨架主体结构；然后提出一种高适应性毛刺去除算法，以分支像素点数量与细化后骨架图像目标像素点数量之比作为判断标准，可以高效适应不同目标像素点密度和规模的裂缝图像。实验结果表明，该算法能够有效实现单一像素宽度骨架并尽可能去除骨架毛刺，有一定的可行性及优越性。

Algorithm of crack skeleton extraction based on template matching and high adaptation deburring

When extracting the skeleton of the crack image, the existing algorithms usually had problems such as the missing information of the skeleton body after the refinement, and the burr removal effect decreased rapidly with the increase of the image size. Aiming at the above problems, this algorithm proposed a crack matching algorithm for template matching and high adaptive burr removal. Firstly, it improved the Rosenfeld refinement algorithm by combining template matching to preserve the skeleton structure. Then proposed a high adaptive burr removal algorithm. The ratio of the number of branch pixels to the number of target pixel points of the refined skeleton image was used as the criterion to effectively adapt to the crack image of different target pixel density and scale. The experimental results show that the algorithm can effectively realize the single pixel width skeleton and remove the redundant skeleton burrs as much as possible, which has certain feasibility and superiority.