为探究高光谱成像技术对羊肉新鲜度无损检测的可行性,通过高光谱成像系统获取羊肉样本935~2 539 nm的高光谱图像,测定羊肉挥发性盐基氮(total volatile basic nitrogen, TVB-N)含量并划分样本新鲜度类别。借助连续投影法(successive projection algorithm,SPA)优选的12个特征波长建立基于反向人工神经网络(back-propagation artificial neural network, BPANN)和分类回归决策树(classification and regression trees , CART)算法的羊肉新鲜度判别模型。结果表明,BPANN模型对校正集和预测集的平均分类准确率为100%和83.33%,对3个新鲜度类别样本的识别率分别为88.89%、75%和85.71%;CART模型对校正集和预测集的平均分类准确率为100%和91.67%,对3个新鲜度类别样本的识别率分别为88.89%、87.50%和100%。CART模型的平均分类准确率和对3个类别样本的识别率均高于BPANN模型,表明高光谱成像技术结合CART算法可有效提高羊肉新鲜度的判别精度。
In order to explore the feasibility of non-destructive detection for lamb freshness by hyperspectral imaging technology, hyperspectral images of lamb samples in the near infrared(935-2 539 nm)regions were acquired by the hyperspectral imaging system, and the total volatile basic nitrogen (TVB-N) for lamb was determined to classify sample freshness. Lamb freshness discrimination model based on back-propagation artificial neural network (BPANN) and classification and regression trees (CART) with 12 characteristic wavelengths optimized by successive projection algorithm (SPA) method. The results showed that the average classification accuracy of BPANN model were 100% and 83.33% for correction set and prediction set, respectively; the recognition rate to three freshness samples was 88.89%, 75% and 85.71%, respectively. The average classification accuracy of CART model were 100% and 91.67% for correction set and prediction set, respectively; the recognition rate to three freshness samples was 88.89%, 87.50% and 100%, respectively. The average classification accuracy and recognition rate of three kinds of samples of cart model were higher than that of BPANN model. The research indicates that hyperspectral imaging technology combined with cart algorithm could effectively improve the accuracy of judging lamb freshness.
[1] 张凯华, 臧明伍,王守伟, 等. 基于光谱技术的畜禽肉新鲜度评价方法研究进展[J]. 肉类研究, 2016, 30(1): 30-35.
[2] FUNAZAKI N, HEMMI A, ITO S, et al.Application of semiconductor gas sensor to quality control of meat freshness in food Industry[J].Sensors and Actuators B (Chemical),1995,25(1-3):797-800.
[3] RUSSELL S,WALLER J.The effect of evisceration on visible contamination and the microbiological profile of fresh broiler chicken carcasses using the Nu-Tech evisceration system or the conventional streamlined inspection system[J]. Poultry Science, 1997, 76(5):780-784.
[4] 栗绍文,包华君,孟宪荣, 等.过氧化物酶反应试纸法检验肉新鲜度试验[J].中国兽医杂志,2003,39(10): 46-47.
[5] 彭彦昆, 张雷蕾.农畜产品品质安全光学无损检测技术的进展和趋势[J].食品安全质量检测学报,2012,3(6):560-568.
[6] CRICHTON S O, KIRCHNER S M, PORLEY V, et al. High pH thresholding of beef with VNIR hyperspectral imaging[J]. Meat Science, 2017, 134:14-17.
[7] CRICHTON S O, KIRCHNER S M, PORLEY V, et al. Classification of organic beef freshness using VNIR hyperspectral imaging[J]. Meat Science, 2017, 129: 20-27.
[8] HE Hongju, SUN Dawen, WU Di. Rapid and real-time prediction of lactic acid bacteria (LAB) in farmed salmon flesh using near-infrared (NIR) hyperspectral imaging combined with chemometric analysis[J]. Food Research International, 2014, 62:476-483.
[9] HE Hongju, SUN Dawen. Toward enhancement in prediction of pseudomonas counts distribution in salmon fillets using NIR hyperspectral imaging[J]. LWT-Food Science and Technology, 2015, 62(1):11-18.
[10] BARBIN D F, ELMASRY G, SUN D W, et al.Non-destructive assessment of microbial contamination in porcine meat using NIR hyperspectral imaging [J]. Innovative Food Science&Emerging Technologies, 2013, 17: 180-191.
[11] GB/5009.228—2016 食品国家安全标准-食品中挥发性盐基氮的测定[S]. 北京: 中国标准出版社, 2016.
[12] GB/2717—2016 食品国家安全标准-鲜(冻)畜、禽产品[S]. 北京: 中国标准出版社, 2016.
[13] 彭彦昆, 张雷蕾. 光谱技术在生鲜肉品质安全快速检测的研究进展[J]. 食品安全质量检测学报, 2010, 27 (2): 14-24.
[14] 张珏, 田海清, 赵志宇, 等. 基于改进离散粒子群算法的青贮玉米原料含水率高光谱检测[J]. 农业工程学报, 2019, 35(1): 285-293.
[15] 王笑丹, 刘爱阳, 孙永海, 等. 基于自组织神经网络模型与质构特性的牛肉嫩度评定方法[J]. 农业工程学报, 2015, 31(18): 262-268.
[16] 郭培源, 徐盼, 董小栋, 等. 高光谱技术结合迭代决策树的香肠菌落总数预测[J]. 食品科学, 2019, 40(6): 312-317.
[17] 杨荣, 赵娟娟, 贾郭军. 基于决策树的存量客户流失预警模型[J]. 首都师范大学学报, 2019, 40(5): 14-18.
[18] 林丽群, 舒宁, 肖俊, 等. 基于遗传算法优化决策树的多光谱影像分类研究[J]. 测绘科学, 2009, 34(4): 122-124.
[19] 黄建琼, 郭文龙, 李秋缘, 等. 基于决策树的城市环境空气质量评价模型实证研究[J]. 科技和产业, 2019, 19(9): 104-110.
[20] 孙梦婷, 魏海平, 李星滢, 等. 利用 CART 分类树分类检测交通拥堵点[J/OL]. 武汉大学学报(信息科学版),https://doi.org/10.13203/j.whugis20190288.
[21] LEE H, KIM M S, LEE W H, et al. Determination of the total volatile basic nitrogen (TVB-N) content inpork meat using hyperspectral fluorescence imaging[J]. Sensors and Actuators B, 2018, 259: 532-539.
[22] 刘思伽, 田有文, 张芳, 等. 采用二次连续投影法和BP人工神经网络的寒富苹果病害高光谱图像无损检测[J]. 食品科学, 2017, 38(8): 277-282.
[23] YU Xinjie, WANG Jianping, WEN Shiting, et al. A deep learning based feature extraction methodon hyperspectral images for nondestructive prediction of TVB-N content in Pacific white shrimp(Litopenaeus vannamei)[J].Biosystems Engineering, 2019, 178: 244-255.
[24] BARBIN D F, ELMASRY G, SUN Dawen, et al. Non-destructive determination of chemical composition in intact and minced pork using near-infrared hyperspectral imaging[J]. Food Chemistry, 2013, 138: 1 162-1 171.
[25] BARBIN D F, ELMASRY G, SUN Dawen, et al. Near-infrared hyperspectral imaging for grading and classification of pork[J]. Meat Science, 2012, 90: 259-268.
[26] KAMRUZZAMAN M, BARBIN D F, ELMASRY G, et al. Potential of hyperspectral imaging and pattern recognition for categorization and authentication of red meat[J]. Innovative Food Science and Emerging Technologies, 2012,16:316-325.
[27] KAMRUZZAMAN M, ELMASRY G, SUN Dawen. Non-destructive assessment of instrumental and sensory tenderness of lamb meat using NIR hyperspectral imaging[J]. Food Chemistry, 2013, 141: 389-396.
[28] TALENS P, MORA L, MORSY N, et al. Prediction of water and protein contents and quality classification of Spanish cooked ham using NIR hyperspectral imaging[J]. Journal of Food Engineering, 2013, 117(3):272-280.
[29] KAMRUZZAMAN M, ELMASRY G, SUN Dawen, et al. Prediction of some quality attributes of lamb meat using near-infrared hyperspectral imaging and multivariate analysis[J]. Analytica Chimica Acta, 2012, 714: 57-67.
[30] PU Hongbin, SUN Dawen, MA Ji, et al. Hierarchical variable selection for predicting chemical constituents in lamb meats using hyperspectral imaging[J]. Journal of Food Engineering, 2014, 143: 44-52.
[31] 刘晓娜, 封志明, 姜鲁光. 基于决策树分类的橡胶林地遥感识别[J]. 农业工程学报, 2013, 29(24): 163-172.
[32] 范中建, 朱荣光, 张凡凡. 基于BP和Adaboost-BP神经网络的羊肉新鲜度高光谱定性分析[J]. 新疆农业科学, 2018, 55(1): 183-188.