基于局部大津阈值与区域生长的坝面细小裂缝识别分割算法

针对传统图像分割算法难以分割噪声污染严重的混凝土坝面细小裂缝图片的问题，结合大津阈值算法和区域生长算法提出一种新的裂缝分割算法。该算法通过一个滑动的窗口遍历整幅灰度图并计算窗口内的局部大津阈值，遍历过程中将所有灰度值在局部阈值以上的窗口中心像素点作为区域生长种子点，设计生长阈值，生长完成后得到分割结果。运用该算法、基于全局的大津阈值等经典分割算法对四幅存在光照不均、污渍覆盖等噪声污染的裂缝图片进行分割。结果表明，所提算法的分割结果中背景噪声最少，误差最小。通过连通域分析去除剩余的背景噪声，可实现对裂缝的精确识别分割，为自动化检测坝面裂缝奠定基础。

Identification and Segmentation Algorithm of Small Cracks on Dam Surface Based on Local OTSU Threshold and Regional Growth

Aiming at the problem that the traditional image segmentation algorithm is difficult to segment the small crack image with serious noise pollution, a new crack segmentation algorithm is proposed, which combines the OTSU algorithm and the region growing algorithm. The algorithm traverses the entire grayscale image through a sliding window and calculates the local OTSU threshold in the window. In the traversal process, all the pixels in the center of the window whose grayscale value is above the local threshold are regarded as the regional growth seed points. Then the growth threshold is designed. Finally, the segmentation results are obtained after the growth is completed. The algorithm presented in this paper and other classical segmentation algorithm like global OTSU algorithm are used to segment four cracks with noise pollution such as uneven illumination, stain coverage, etc. The segmentation results of the algorithm presented in this paper have the least background noise and the least error. The remaining background noise is removed through connected domain analysis, and the accurate identification and segmentation of cracks are realized, which lays the foundation for automatic crack monitoring.