基于计算机视觉的爆破块度分布统计

Statistics of Blasting Fragmentation Distribution Based on Computer Vision

爆破块度是评价矿山爆破效果的重要指标。传统的块度统计方法存在效率低、限制多等问题。基于计算机视觉的爆堆块度统计方法,具有高效、精确、灵活的优势。针对爆堆块度统计的计算机视觉识别问题,提出了一种结合Mask B-R-CNN和HSV变换的形态学优化的块度统计方法,并对实验室和鞍千矿业公司的矿山现场爆堆块度统计进行了验证。结果表明,与筛分法相比,基于计算机视觉的爆堆块度分布统计方法的误差小于3%,验证了该方法用于爆堆块度统计的可行性。矿山现场试验结果表明,与传统的矿岩区域提取方法相比,该方法提取矿岩区域更为精准,可应用于鞍千矿业公司的矿山现场;3个爆堆块度分布的累计概率曲线相似,大块率分别为4.21%、3.37%、3.12%,爆破效果较好。

Blasting fragmentation is an important index to evaluate the blasting effect of mines.The traditional fragmentation statistical method has the problems of low efficiency and many restrictions.The statistical method of blasting fragmentation based on computer vision has the advantages of high efficiency,accuracy and flexibility.Aiming at the problem of computer vision recognition of blasting piles fragmentation statistics,a morphologically optimized method of fragmentation statistical combining Mask B-R-CNN and HSV transform was proposed,and the blasting piles fragmentation statistics of the laboratory and the mine site of Anqian Mining Company were verified.The results show that compared with the screening method,the error of fragmentation distribution statistical method based on computer vision is less than 3%,which verifies the feasibility of this method for the blasting piles fragmentation statistics.The field experiment results of mine show that compared with the traditional method of extracting ore and rock area,this method is more accurate in extracting ore and rock area,and can be applied to the mine site of Anqian Mining Company.The cumulative probability curves of the three blasting fragmentation distributions are similar,and the large fragmentation rates are 4.21%,3.37%and 3.12%respectively,and the blasting effect is better.