摘要
针对阴极铜板生产过程采用人工目测表面结瘤存在识别效率低、劳动量大和不可连续等问题,提出一种基于改进灰狼优化(IGWO)算法的阴极铜板表面结瘤智能检测方法。采用最大熵(ME)准则构建目标函数,引入灰狼优化(GWO)算法在可行域内全局搜索铜板图像的结瘤分割阈值;利用混沌反向学习和自适应T分布改进GWO,确保其多样性不随种群进化逐代降低、协助算法跳出局部最优解,增加最佳阈值的寻取机率;同时提出掩膜扫描法消除铜板图像冗余纹理信息,增加结瘤面积比识别准确率;最后通过识别结果智能判别阴极铜板品味。实验结果表明:IGWO-ME法较5种智能算法对阴极铜板图像的结瘤分割,峰值信噪比(PSNR)、结构相似性(SSIM)整体分别提升0.0198~0.1143 dB、0.00016~0.03465;较ME、Otsu和IGWO-Otsu在分割时耗上至少提升56.1%;检验结果表明该方法对阴极铜板表面结瘤检测具备可行性和有效性。
In response to the problems of low identification efficiency,high labor and non-continuity in the production process of copper cathode plate using manual visual inspection of surface nodules,an intelligent detection method of surface nodules on copper cathode plate based on the Improved Gray Wolf Optimization(IGWO)algorithm is proposed.Firstly,the maximum entropy(ME)criterion is used to construct the objective function,and the grey wolf optimization(GWO)algorithm is introduced to globally search for the tumor segmentation threshold of copper plate images in the feasible domain;then,chaotic backward learning and adaptive T-distribution are used to improve GWO to ensure that its diversity does not decrease generation by generation with the evolution of the population,assist the algorithm to jump out of the local optimal solution,and increase the chance of finding the best threshold;meanwhile,the mask scanning method is proposed to eliminate.At the same time,the mask scanning method is proposed to eliminate the redundant texture information of copper plate image and increase the recognition accuracy of nodule area ratio;finally,the recognition results are used to intelligently discriminate the taste of cathode copper plate.The experimental results show that the IGWO-ME method improves PSNR and SSIM by 0.0198~0.1143 dB and 0.00016~0.03465 respectively compared with the five intelligent algorithms for nodule segmentation of cathode copper plate images,and improves the segmentation time consumption by at least 56.1%compared with ME,Otsu and IGWO-Otsu;the validity test results show that the method is feasible for cathode.The validity test results show that the method is feasible and effective for the detection of surface nodules on copper plates.
作者
杨雪松
唐都作
蔡兵
袁海滨
潘新远
常以涛
张长胜
YANG Xuesong;TANG Duzuo;CAI Bing;YUAN Haibin;PAN Xinyuan;CHANG Yitao;ZHANG Changsheng(Faculty of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650504,China;Yunnan Tin Group(Holding)Co,Ltd,Honghe 661400,China)
出处
《陕西理工大学学报(自然科学版)》
2023年第3期31-39,共9页
Journal of Shaanxi University of Technology:Natural Science Edition
基金
国家自然科学基金项目(51665025,61963022)。
关键词
阴极铜板结瘤
最大熵
灰狼优化算法
混沌反向学习策略
自适性T分布扰动
the cathode copper plate nodules
maximum entropy
gray wolf optimization algorithm
chaotic opposition learning strategy
adaptive T-distribution disturbance