基于手机图像的不同贮藏时间下冷却羊肉的部位判别  

Discrimination of Chilled Lamb from Different Carcass Parts at Different Storage Times Based on Mobile Phone Images

作　　者：孟令峰 朱荣光[1] 白宗秀 郑敏冲 顾剑峰 马本学[1] MENG Lingfeng;ZHU Rongguang;BAI Zongxiu;ZHENG Minchong;GU Jianfeng;MA Benxue(College of Mechanical and Electrical Engineering,Shihezi University,Shiheizi 832003,China)

机构地区：[1]石河子大学机械电气工程学院,新疆石河子832003

出　　处：《食品科学》2020年第23期21-26,共6页Food Science

基　　金：国家自然科学基金地区科学基金项目(31860465,31460418);兵团中青年科技创新领军人才计划项目(2020CB016);石河子大学青年创新人才培育计划项目(CXRC201707)。

摘　　要：为了实现手机对冷却羊肉不同贮藏时间下不同部位的快速判别,本研究利用手机采集不同贮藏时间(0~12 d)下羊背脊肉、羊前腿肉和羊后腿肉样品的图像,提取不同颜色空间下的颜色均值和RGB颜色空间下的颜色矩等颜色特征并进行差异显著性分析,获得不同部位之间具有显著性差异的7个颜色特征。根据不同部位之间颜色特征的差异性以及不同的颜色空间,选定4种颜色特征组合作为模型输入,分别利用极限学习机(extreme learning machine,ELM)、支持向量机(support vector machine,SVM)和反向传播(back propagation,BP)神经网络进行羊背脊肉、羊后腿肉和羊前腿肉的分类比较研究。结果表明:以不同的颜色特征组合作为模型输入时,所建立的BP模型分类效果均优于SVM和ELM模型;当以12个颜色均值特征作为输入时所建立的BP模型分类效果最优,该模型的训练集、交叉验证集和测试集的判别准确率分别为96.13%、95.11%、91.44%,可以实现对不同贮藏时间下不同部位羊肉的定性判别分析。上述研究为后续开发手机应用APP及利用手机实现对不同贮藏时间下冷却羊肉部位的快速判别分析提供了理论依据和技术支撑。In order to realize the rapid identification of different cuts of chilled lamb at different storage times by using mobile phones,images of back,hind leg and front leg were collected over a storage time of 12 days.Color features such as the mean value in different color spaces and the color moments in the red,green and blue(RGB)color space were extracted and analyzed for significant differences,and 7 color features with significant differences among meat cuts were obtained.Considering the differences in color features among meat cuts as well as different color spaces,four combinations of color features were selected as the inputs for the extreme learning machine(ELM),back propagation(BP)and support vector machine(SVM)models,respectively,and modeling analysis and comparison were carried out to classify different lamb cut.The results showed that upon using each color feature combination as the model input,the classification performance of the BP model established was better than that of the SVM and ELM models.When the 12 color mean features were chosen as the input,the BP model showed the best classification performance with a discrimination accuracy of 96.13%,95.11%and 91.44%for the training,cross-validation,and test set,respectively,allowing rapid discrimination of lamb cuts at different storage times.The research provides a theoretical basis and technical support for future development of mobile phone applications to rapidly discriminate chilled lamb cuts at different storage times.

关 键 词：冷却羊肉 不同部位 手机图像 贮藏时间 

分 类 号：TS251.7[轻工技术与工程—农产品加工及贮藏工程]