基于计算机视觉对苹果脆片外观品质分级

发布时间：2018-12-08 12:15

【摘要】：[目的]为了开发苹果脆片自动化分级技术,利用计算机视觉对苹果脆片外部品质进行无损检测研究。[方法]通过对苹果脆片的大小、形状、颜色、纹理特征的感官评分建立苹果脆片外部品质等级评价标准。首先,利用计算机视觉系统获取脆片图像,经图像处理后,进行特征提取;然后,通过比较数据离差标准化和标准差标准化处理对费歇尔线性判别(FLD)、偏最小二乘法判别分析(PLS-DA)、支持向量机(SVM)3种模型建模效果的影响,确定合适的数据预处理方式;并分别基于这3种模型对预测样本进行分级;最后,采用连续投影算法(SPA)筛选最佳特征参数,并比较3种模型的建模和预测结果,确定最佳分级模型。[结果]数据经离差标准化处理后建模效果更好;FLD、PLS-DA、SVM 3种模型中SVM模型分级效果最好,其建模集和预测集分级准确率分别为99.1%和98.0%;基于SPA算法筛选得到了面积、圆形度、对比度的均值、能量的均值、对比度的均方差、熵的均方差这6个特征参数,基于特征参数的SVM模型建模集和预测集分级准确率分别为98.7%和97.3%。[结论]该研究实现了苹果脆片的外观品质的自动检测和分级,为苹果脆片的计算机视觉自动分级提供了技术支持。
[Abstract]:[objective] in order to develop the automatic grading technology of apple chips, the external quality of apple chips was studied by computer vision. [methods] the external quality evaluation criteria of apple chips were established by sensory scoring of the size, shape, color and texture features of apple chips. Firstly, the embrittlement image is obtained by computer vision system, and the feature is extracted after image processing. Then, the effects of data deviation standardization and standard deviation standardization on Fisher linear discriminant (FLD), partial least squares discriminant analysis (PLS-DA) and support vector machine (SVM) (SVM) modeling are compared. To determine the appropriate data preprocessing mode; The prediction samples are classified based on the three models. Finally, the optimal feature parameters are screened by continuous projection algorithm (SPA), and the results of modeling and prediction of the three models are compared to determine the best classification model. [results] the modeling effect of the data was better after the deviation standardization, among the three FLD,PLS-DA,SVM models, the SVM model had the best classification effect, and the classification accuracy of the modeling set and the prediction set were 99.1% and 98.0%, respectively. Based on the SPA algorithm, six characteristic parameters are obtained: mean value of area, roundness, contrast, mean energy, mean variance of contrast, mean variance of entropy. The classification accuracy of modeling set and prediction set of SVM model based on feature parameters is 98.7% and 97.3% respectively. [conclusion] this study has realized the automatic inspection and grading of the appearance quality of apple chips, and provided the technical support for the automatic grading of apple chips by computer vision.
【作者单位】：南京农业大学食品科技学院;江苏派乐滋食品有限公司;
【基金】：国家自然科学基金项目(31671925,31671926) 苏北科技发展计划——科技富民强县项目(BN2015025) 中央高校基本科研业务费专项资金(KYLH201504)
【分类号】：TP391.41;TS255.4

【相似文献】