基于PSO进化神经计算进行苹果颜色快速分级.pdfVIP

1-1777／TP 计算机研究与发展 ISSN1000—1239／CN of Researchand 42(增刊)：263～268，2005 JournalComputer Development 基于PSO进化神经计算进行苹果颜色快速分级 郭志涛 夏克文 袁金丽 (河北工业大学信息工程学院天津300130) (mrnow@jsmail．hebut．edu．cn) Color at BasedonPSO Neural Apple GradingHighSpeed Evolutionary Computation Yuan Guo Jinli Zhitao，XiaKewen，and (School 300130) ofInformationEngineering，HebeiUniversityofTechnology，Tia巧in AbstractInordertoeliminatethe in color asslow error， shortcomingsapple grading，suchspeed，large etc．，anovelfast methodis isbased0nthe swarm intelligentgrading presented，which particle Optimization swarm andtheneural main (PSO)algorithmreappearingintelligence computationtechnology．Theprocess isto the of surfacethe vision andextractits acquirecolorityapple by computertechnology features，then traintheneuralnetworkarchitecturesan PSO the colorwith by improvedalgorithm，andfinallygradeapple thetrainednetwork．Theactual showsthatthemethodcanachieve obtain application highprecision，and fast color effectis notable． very gradingspeed．Inapple grading，theapplicationvery words swarm color Key particle optimizationalgorithm；neuralcomputing；applegrading 摘要为了克服在苹果颜色分级中存在的速度慢、误差大等缺点，基于再现群智能的粒子群优化算法 和神经计算技术，提出了一种新颖、快速的智能分级方法，即首先通过计算机视觉技术获取苹果表面颜 色的色度，并提取其特征；然后采用改进的粒子群优化算法训练神经网络结构；最后用训练好的神经网 络进行苹果颜色分级．实际应用表明，该方法切实可行且效果显著，不仅分级速度快，而且分级正确率 高达96％以上． 关键词粒子群优化算