中国农村水利水电Issue(1):184-189,6.DOI:10.12396/znsd.2500199
多尺度视觉测量融合的堆石料表观级配检测方法
A Method for Identifying Apparent Rockfill Gradation via Integrated Multi-scale Visual Measurement
摘要
Abstract
Due to the significant scale differences in rockfill gradation,common methods for image segmentation on particles often suffer from over-segmentation or under-segmentation,struggling to meet the requirements of engineering applications.Therefore,this paper proposes a contour extraction method integrating multi-scale visual measurement.Firstly,images of the paved rockfill are collected.Then,to address the problem of uneven segmentation scales caused by large particle size spans of rockfill,an integrated multi-scale visual measurement model is constructed in combination with Mask R-CNN,presenting an accurate approach to extract the contours of stones in different sizes.Based on the extracted results,a model is established to reconstruct the three-dimensional shape of particles and calculate the gradation,enabling fast evaluation of the apparent gradation of rockfill materials.Field experiments exhibit that this method provides high-precision identification of large-span particle sizes and can effectively support quick assessment of rockfill gradation on site.关键词
堆石料/级配检测/图像分割/信息融合/特征提取Key words
rockfill/gradation detection/image segmentation/information integration/feature extraction分类
建筑与水利引用本文复制引用
薛守宁,喻葭临,周宜红,赵春菊,段斌,覃事河,李祥,冯德强..多尺度视觉测量融合的堆石料表观级配检测方法[J].中国农村水利水电,2026,(1):184-189,6.基金项目
国家能源集团科技创新项目资助(GJNY-DDH-2023-014). (GJNY-DDH-2023-014)