基于机器视觉的甜瓜成熟度判别研究  

Study on the Maturity Discrimination of Muskmelon Based on Machine Vision

作　　者：许德芳 赵华民[1] 许建东 张淑娟[1] XU Defang;ZHAO Huamin;XU Jiandong;ZHANG Shujuan(College of Engineering,Shanxi Agricultural University,Jinzhong,Shanxi 030801,China)

机构地区：[1]山西农业大学工学院,山西晋中030801

出　　处：《农产品加工》2020年第14期19-22,25,共5页Farm Products Processing

基　　金：山西省高等学校科技创新项目(2019L0402);山西省优秀博士来晋工作奖励资金科研项目(SXYBKY2018030);博士科研启动项目(2018YJ43)。

摘　　要：为了满足甜瓜自动加工分选的需求,提高甜瓜分拣的效率,采用机器视觉技术对甜瓜进行成熟度判别研究。采集"新甜"甜瓜3个不同成熟度阶段的图像。首先,通过RGB色彩模型和灰度共生矩阵(GLCM)分析、计算甜瓜颜色及纹理特征,然后基于全部特征值和经主成分优选后特征值,建立PLS-LS-SVM预测模型。结果表明,经主成分优选后建立的PCA-LS-SVM预测模型效果最好,综合判别准确率为100%。基于全特征值建立的PLS模型和经主成分优选后建立的PCA-PLS模型综合判别准确率均为97.96%,但是PCA-PLS模型判别速度更高。研究为甜瓜智能化分级分选系统的研发提供了理论支持。In order to meet the needs of melon automatic processing and sorting,and improve the efficiency of muskmelon sorting.In this paper,machine vision technology was used to distinguish the maturity of melon.Firstly,the color and texture characteristics of melon were analyzed and calculated by RGB color model and GLCM.Then,PLS-LS-SVM prediction models were established based on all eigenvalues and eigenvalues optimized by principal components.The experimental results showed that the PCA-LS-SVM model was the best,and the comprehensive discrimination accuracy was 100%.The comprehensive discrimination accuracy of PLS model based on full eigenvalue and PCA-PLS model based on principal component optimization was 97.96%,but PCA-PLS model has higher discrimination speed.This study provided theoretical support for the research and development of muskmelon intelligent classification system.

关 键 词：机器视觉 甜瓜 PCA PCA-LS-SVM 

分 类 号：S233.5[农业科学—农业机械化工程]