基于数字图像处理技术在裂纹检测方面的实现

Realization of Crack Detection Based on Digital Image Processing Technology

摘要: 随着我国高速公路建设的快速发展，高速公路路基路面的质量监控体系日趋完善，对路面检测水平的要求也越来越高。文章提出基于图像处理的裂纹检测方法研究与实现，在图像预处理、图像分割、图像边缘检测方面进行了研究，利用Matlab程序仿真实现了公路裂纹检测，得到清晰可辨的裂纹特征。图像预处理包括：图像增强、图像平滑和图像锐化。其中图像增强采用直方图均衡化和高斯滤波法；图像平滑采用中值滤波法；图像锐化采用Sobel算子法。区别于未处理图像，裂纹图像经预处理后，相较原图像细节更明显。图像分割采用Otsu法阈值分割；边缘检测采用Canny边缘检测。这两种方法能不同于其他图像分割、边缘检测方法，更易体现裂纹图像特征。经数字图像处理后的裂缝图像，具备更易识别、更易发现的优点，更好帮助公路维护人员及时发现问题。

Abstract: With the rapid development of expressway construction in our country, the quality monitoring sys-tem of expressway subgrade and pavement is becoming more and more perfect, and the requirement of pavement inspection level is higher and higher. This paper puts forward the research and implementation of crack detection method based on image processing, and studies image prepro-cessing, image segmentation and image edge detection. Highway crack detection is simulated by Matlab program, and clear and discernible crack features are obtained. Image preprocessing includes image enhancement, image smoothing and image sharpening. Histogram equalization and Gaussian filtering are used for image enhancement, median filtering is used for image smoothing, and Sobel operator method is used for image sharpening. Different from the unprocessed image, the details of the crack image are more obvious than the original image after preprocessing. Otsu threshold segmentation is used for image segmentation, and Canny edge detection is used for edge detection. These two methods can be different from other image segmentation and edge detection methods, and are easier to reflect the characteristics of crack images. The crack image processed by digital image has the advantages of easier identification and discovery, and can better help highway maintenance personnel to find problems in time.