摘要
针对大米加工中的品质智能化识别要求,提出了一种基于图像处理的大米品质识别系统。首先,采用面积法对大米外观进行完整性检测,从而将大米分为完整粒和碎米;其次,在完整粒划分的基础上,采用麻雀搜索算法(SSA)优化支持向量机(SVM)参数,并对完整粒色调和垩白度进行识别,以区分完善粒、垩白粒和黄米粒,进而完成大米品质的识别。实验结果表明,采用面积法的大米大小分级法对米粒完整性划分的准确度达94.3%;在完整粒识别基础上,采用的SSA-SVM分类模型对大米图像色调的识别准确率达98.21%,相较于传统SVM分类提高了1.63个百分点,表现出良好的识别精度;将上述算法部署到识别软件中,可有效对大米品质进行识别划分。由此得出,以上识别方案可行,可用于大米品质加工。
Aiming at the requirement of intelligent quality recognition in rice processing,a rice quality recognition system based on image processing was proposed.Firstly,the area method was used to detect the integrity of the appearance of rice,and then the rice was divided into whole grains and broken rice.Secondly,on the basis of complete grain partitioning,Sparrow Search Algorithm(SSA) was used to optimize the parameters of Support Vector Machine(SVM),and the complete grain hue and chalkiness were identified.To distinguish the perfect grain,chalky grain and yellow rice grain,and then complete the identification of rice quality.The experimental results showed that the accuracy of rice size classification by area method was 94.3%.On the basis of whole grain recognition,the SSA-SVM classification model has a recognition accuracy of 98.21%,which is 1.63% higher than the traditional SVM classification,showing good recognition accuracy.The above algorithms can be deployed in the recognition software to identify and classify rice quality effectively.It is concluded that the above identification scheme is feasible and can be used for rice quality processing.
作者
张健
刘越君
ZHANG Jian;LIU Yue-jun(Hanzhong Vocational and Technical College,Hanzhong 723002,China)
出处
《粮食与饲料工业》
CAS
2024年第3期1-5,共5页
Cereal & Feed Industry
基金
陕西高等职业教育教学改革研究项目(21GY047)
汉中职业技术学院教学研究与改革项目(HZZYJY2021003,HZZYGL2021004)
汉中职业技术学院教学研究与改革项目(HZZYJY2021068)。
关键词
机器学习
图像处理
大米品质
麻雀搜索算法
面积法
machine learning
image processing
food quality
Sparrow search algorithm
area method
