基于GAS算法的卵砾石粒径自动识别应用研究  

Study on the application of automatic identification of gravel particle size based on GAS algorithm

作　　者：蔡豫豪 高仕赵[1] 张丛林 董晓明[1] CAI Yu-hao;GAO Shi-zhao;ZHANG Cong-lin;DONG Xiao-ming(School of Civil Engineering,Ludong University,Yantai 264025,China;Institutes of Science and Development,Chinese Academy of Sciences,Beijing100190,China)

机构地区：[1]鲁东大学土木工程学院,山东烟台264025 [2]中国科学院科技战略咨询研究院,北京100190

出　　处：《泥沙研究》2024年第2期9-16,共8页Journal of Sediment Research

基　　金：山东省中青年专家奖励基金项目(BS2014SF016)。

摘　　要：粒径和级配是表征床面组成的重要指标,基于GAS粒径自动识别技术可自动识别粗粒床面粒径并生成级配曲线,能够大幅提高现场采样和分析的效率。为了验证GAS的分割效果,采用GAS提供的默认参数进行分割,同时应用ImageJ软件手动分割进行验证。结果表明:GAS级配曲线的相对误差为5.7%,相关系数为0.992。另外,采用单参数和多参数敏感性分析法来标准化参数调整方案,gre、can1和can2对GAS提取的级配曲线和特征粒径有显著影响,其中gre起主导作用,而can1和can2控制着砾石边界的检测完整性。Grain size and its distribution are vital indicators for characterizing coarse-grained beds.Based on GAS particle size automatic recognition technology,coarse particle bed surface particle size can be automati-cally identified and grading curves can be generated,greatly improving the efficiency of on-site sampling and analysis.To verify the segmentation effect of GAS,the default parameters provided by GAS are used for seg-mentation,while ImageJ software is applied for manual segmentation for verification.The results show that the relative error of the GAS grading curve is 5.7%,and the correlation coefficient is 0.992.Standardize parame-ter adjustment schemes by single parameter and multi parameter are also conducted to analysis the sensitivity.Parameters of gre,canl and can2 can significantly affect the grading curve and characteristic grain sizes ex-tracted by GAS.Among them,gre plays a dominant role,while canl and can2 control the detection integrity of gravel boundary.

关 键 词：粗粒床面 级配曲线 GAS算法 数字筛分 图像处理 

分 类 号：P332.5[天文地球—水文科学]