煤矿胶带表面裂纹检测图像处理算法研究

Research on Image Processing Algorithm for Surface Crack Detection of Coal Mine Belt

采用分治法改进的中值滤波法进行滤波去噪,提高算法执行速度;使用直方图均衡化与局部对比度结合法增强图像的动态范围,凸显裂纹细节,同时将Sobel算子扩充为四个方向进行边缘检测,仿真结果表明扩充的Sobel算子输出边缘更强烈,更有利于图像分割。提出一种实时性更强的改进二维Otsu算法进行图像分割,从图像实验结果可以看出,改进分割算法效果非常理想、有更好的自适应性,受图像对比幅度变化、亮度变化和目标比例大小的影响较小。

Based on the characteristics of underground belt crack image, the author presents a set of image processing algorithm of crack detection by comparative analysis, in which firstly, the median filtering method is used to filter noise and improve the algorithm execution speed;secondly, histogram equalization and local contrast method are used to enhance the dynamic range of the image, highlight the crack details, and expand the Sobel operator into four directions for edge detection;and finally a more real-time improved two-dimensional Otsu algorithm is proposed for image segmentation with ideal algorithm effects, better self-adaptability and less impacts of the change of contrast amplitude, brightness change and target ratio from the experimental results of the image.