基于改进蝗虫算法优化Canny算子的鸡蛋裂纹图像检测  

Egg crack image detection method based on improved grasshopper optimization algorithm and canny operator

作　　者：涂伟沪 蔡玲霞 李学军[2] TU Wei-hu;CAI Ling-xia;LI Xue-jun(Xinjiang Hami Radio and Television University,Hami,Xinjiang 839000,China;Jincheng School,Sichuan University,Chengdu,Sichuan 611731,China)

机构地区：[1]新疆哈密广播电视大学,新疆哈密839000 [2]四川大学锦城学院,四川成都61173

出　　处：《食品与机械》2022年第2期167-172,202,共7页Food and Machinery

基　　金：四川省科技计划软科学研究项目(编号:2019JDR0030);四川省区域创新合作项目(编号:2020YFQ0018)。

摘　　要：目的:提高鸡蛋线形、网状裂纹检测率。方法:针对蝗虫优化算法(GOA)求解高维复杂优化问题时收敛效率不高的缺陷,设计改进的模糊C-均值算法(FCM)对蝗虫种群进行等级划分;设计自适应极值逆向学习和编码突变更新机制,以扩展算法深度搜索空间和全局寻优能力;采用改进的GOA对参数进行优化,并将改进蝗虫算法优化Canny算子用于鸡蛋裂纹检测。结果:该方法对鸡蛋线形裂纹和网状裂纹漏检率分别降低了21.4%~31.2%,63.2%~69.7%,优于其他算法。结论:该方法能有效提升鸡蛋裂纹检测准确率。Objective: In order to improve the detection effect of egg linear and reticular cracks. Methods: As the low convergence efficiency of Grasshopper optimization algorithm(GOA) in solving high-dimensional complex optimization problems, an improved fuzzy c-means algorithm(FCM) was designed to classify locust population. The adaptive extreme value reverse learning and coding mutation update mechanism were designed to expand the depth search space and global optimization ability of the algorithm. The improved GOA was used to optimize the parameters, and the improved canny operator was used for egg crack detection. Results: The results showed that the missed detection rates of egg linear crack and mesh crack were improved by about 21.4%~31.2% and 63.2%~69.7% respectively, which was better than other algorithms. Conclusion: This method can effectively improve the accuracy of egg crack detection.

关 键 词：鸡蛋 裂纹 蝗虫优化算法 模糊C-均值 漏检率 

分 类 号：TS253.7[轻工技术与工程—农产品加工及贮藏工程] TP391.41[轻工技术与工程—食品科学与工程] TP18[自动化与计算机技术—计算机应用技术]