基于双门限阈值的爆破块度图像识别研究

为改进爆破块度图像分割中容易存在的过分割、噪声、"黑洞"现象等问题,提出了双门限阈值分割技术,开展了多种类型的爆破岩块图像分割实验和小型爆堆的手工测量实验,实验结果表明,双门限阈值法与灰度阈值法、适应性阈值法、Otsu大津法等常用的分割方法相比,对爆破岩堆的图像分割效果更优,能够解决岩石表面噪声问题;与手工测量相比,双门限阈值技术对爆堆图像分割的平均相对误差为12.0%,误差主要因岩块的边缘棱角效应所致,这种效应使得分割结果偏大。双门限阈值分割技术能较好地应用于爆破块度图像识别。

Research on image recognition of blasting block based on double threshold

In order to improve the problems of over segmentation, noise and "black hole" that are easy existing in image segmentation of blasting rock mass, the double threshold segmentation technology is proposed, and many kinds of experiments on image segmentation of blasting rock mass and manual measurement of small blasting piles are carried out. The experimental results show that the double threshold method compare with the commonly used segmentation methods such as gray threshold method, adaptive threshold method, Otsu method, etc is more effective in image segmentation of blasting rock pile, and it can solve the problem of rock surface noise; compared with the manual measurement, the average relative error of the double threshold technology in image segmentation of blasting rock pile is 12.0%.The error is mainly caused by the edge angular effect of rock block, which makes the segmentation result larger. The double threshold segmentation technology can be applied to the image recognition of blasting fragmentation.