基于多颜色空间的太阳能电池片智能分类  

Intelligent Classification of Solar Panels Based on Multi-Color Space

作　　者：王倩 高向军 葛方振[1] 沈龙凤[1] 李想 刘怀愚[1] WANG Qian;GAO Xiangjun;GE Fangzhen;SHEN Longfeng;LI Xiang;LIU Huaiyu(College of Computer Science and Technology,Huaibei Normal University,Huaibei 235000)

机构地区：[1]淮北师范大学计算机科学与技术学院,淮北235000

出　　处：《计算机与数字工程》2021年第3期556-561,共6页Computer & Digital Engineering

基　　金：安徽省重点研究与开发计划面上攻关项目(编号:201904a05020072);安徽省自然科学基金项目(编号:1808085MF174,1808085QF181);安徽高校自然科学研究项目(编号:KJ2019A0606,KJ2019A0603)资助。

摘　　要：针对目前人工分类太阳能电池片效率低、误差大的问题,提出了一种基于太阳能电池片多颜色空间信息的LeNet-5智能分类模型。首先构建太阳能电池片基本分类模型,优化LeNet-5网络结构提高模型性能;然后分析不同颜色空间在太阳能电池片分类中的作用,并给出多颜色空间分类融合算法。实验结果表明,RGB+Lab+HSV的三种颜色空间组合模型分类效果最佳,准确率高达94.56%,基本达到工业应用要求。To solve the problem of low efficiency and large error in manual classification of solar panels,an intelligent LeNet-5 classification model based on multi-color space information of solar panels is proposed.Firstly,the basic classification model of solar panels is constructed,and the LeNet-5 network structure is optimized to improve the performance of the model.Then,the role of different color spaces in the classification of solar panels is analyzed,and a multi-color space classification fusion algorithm is given.The experimental results show that RGB+Lab+HSV three color space combination models have the best classifica⁃tion effect,and the accuracy is 94.56%,which basically meets the requirements of industrial applications.

关 键 词：太阳能电池片 LeNet-5 颜色空间 图像分类 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]