基于IGS-SVM模型的牛肉生理成熟度预测方法研究

摘要/Abstract

摘要： 摘要：生理成熟度是判定牛肉质量等级的重要指标，提出一种通过改进的网格搜索（Improved Grid Search,IGS）算法优化支持向量机（Support Vector Machine,SVM）参数的模型，实现牛肉的生理成熟度预测的方法。收集了18、36、54以及72个月龄的牛肉样本各25个，共计100个。利用机器视觉，采集样本的显微图像，经过图像处理后，提取不同生理成熟度牛肉的肌纤维特征参数，用统计学方法分析牛肉生理成熟度和肌纤维特征参数之间的相关性，并以肌纤维特征参数作为输入，利用76个训练集样本，建立牛肉生理成熟度的SVM预测模型。为优化所建立SVM模型，提出一种改进的网格搜索算法，用其对SVM模型的约束参数C以及核函数参数g进行优化，并结合留一交叉验证法（Leave-One-Out Cross Validation,LOO-CV）得到最优的（C，g）参数组合，将最佳参数代入分类器，得到优化的牛肉生理成熟度预测模型。用24个测试集的独立样本检测模型的适用性和估测性能，结果表明，利用该模型对牛肉的生理成熟度预测的准确率可达到91.67%；与传统网格搜索算法（Grid Search,GS）相比，IGS算法使得模型在训练时间上减少了1755.41s。这表明所建立的模型具有较好的预测效果，也说明根据牛肉肌纤维的特征参数结合机器视觉及图像处理技术，对牛肉生理成熟度进行自动判定的方法是可行的。

关键词: 关键词：牛肉, 生理成熟度, 支持向量机, 预测模型

Abstract: Abstract: Physiological maturity is an important indicator to determine the quality level of beef. This paper proposes a method to predict the physiological maturity of beef by optimizing the model of support vector machine (SVM) parameters through improved grid search (IGS) algorithm. A total of 25 beef samples of 18, 36, 54 and 72 months old were collected, totaling 100. Using machine vision, the microscopic images of the samples were collected. After image processing, the muscle fiber characteristic parameters of beef with different physiological maturity were extracted, and the correlation between the physiological maturity of beef and the characteristic parameters of muscle fibers was analyzed by statistical methods. Using the muscle fiber characteristic parameters as input, 76 training set samples were used to establish a SVM prediction model for beef physiological maturity. In order to optimize the SVM model which has been established, an improved grid search algorithm is proposed, which is used to optimize the constraint parameter C and kernel function parameter g of the SVM model. Combined with the leave-one-out cross validation method (LOO-CV), the optimal (C, g) parameter combination was obtained, and the optimal parameters were substituted into the classifier to obtain an optimized prediction model of beef physiological maturity. The applicability and estimation performance of the beef physiological maturity prediction model were tested with independent test samples of 24 test sets. The results showed that the accuracy of prediction of the physiological maturity of beef using the model could reach 91.67%. Compared with the traditional grid search algorithm (GS), the IGS algorithm reduces the model training time by 1781.3s. There is a significant correlation between beef muscle fiber characteristics and cattle age; according to the characteristic parameters of beef muscle fiber combined with machine vision technology, the physiological maturity of beef could be determined automatically.

Key words: Keywords:beef, physiological maturity, Support Vector Machine（SVM）, prediction model

中图分类号:

TS251.1

季方芳吴明清赵阳陈坤杰. 基于IGS-SVM模型的牛肉生理成熟度预测方法研究[J]. 食品科学, 0, (): 0-0.

Fang fang JI. Prediction model of beef physiological maturity based on IGS-SVM[J]. FOOD SCIENCE, 0, (): 0-0.

参考文献

[1]汤晓艳, 周光宏, 徐幸莲.大理石花纹、生理成熟度对牛肉品质的影响[J].食品科学, 2006, 27(12):114-117
[2]李岩, 刘波, 陈怡帆.机器视觉技术在牛肉生理成熟度检测中的应用[J].河北农业大学学报, 2017, 40(06):119-124
[3]潘超, 彭增起, 周光宏.中国黄牛生理成熟度和大理石花纹预测[J].畜牧与兽医, 2006, 38(10):32-34
[4]李春保.中国黄牛生理成熟度研究[D]. 南京：南京农业大学, 2003.
[5]KRANZLER G A.Beef quality grading using machine vision[J]. [J].Proceedings of the SPIE, 2000, 4203(12):91-101
[6]JEYAMKONDAN S, RAY N, KRANZLER G A, et al.Beef quality grading using machine vision[C]// Environmental and Industrial Sensing. International Society for Optics and Photonics, 2000.
[7]SUN X, CHEN K J, BERG E P, et al.Predicting Fresh Beef Color Grade Using Machine Vision Imaging and Support Vector Machine (SVM) Analysis[J].Journal of Animal & Veterinary Advances, 2011, 10(12):1504-1511
[8]陈坤杰.基于分形理论及机器视觉的牛肉自动分级技术研究[D]. 南京：南京农业大学, 2005.
[9]陈坤杰, 尹文庆.机器视觉技术在分析牛肉颜色变化特征中的应用[J].食品科学, 2008, 29(9):92-96
[10]胡宝利.不同年龄秦川牛胴体性状与肉质性状的研究[D].咸阳：西北农林科技大学, 2001.
[11]郎玉苗.肌纤维类型对牛肉嫩度的影响机制研究[D]. 北京：中国农业科学院, 2016.
[12]章杰, 李学伟, 勇立.肌纤维与肉质的关系[J]. [J].猪业科学, 2013, 3(3):116-118
[13]谌启亮, 彭增起, 沈明霞, 等.牛肉肌纤维直径和结缔组织含量与嫩度相关性研究[J].食品科学, 2012, 33(13):126-129
[14]刘苏苏, 孙立民.支持向量机与神经网络回归性能比较研究[J].计算机工程与设计, 2011, 32(12):4202-4205
[15]刘木华, 段武貌, 黎静, 等.基于图像处理和支持向量机分类的牛肉大理石花纹等级评定[J].沈阳农业大学学报, 2005, 36(6):650-654
[16]吴一全, 曹鹏祥, 王凯, 等.基于不变矩和改进SVM的牛肉大理石纹评级[J]. [J].现代食品科技, 2015, 4(4):17-22
[17]魏丽冉, 岳峻, 李振波, 等.基于核函数支持向量机的植物叶部病害多分类检测方法[J].农业机械学报, 2017, 48(S1):166-171
[18]SUN X, ChEN K J, BERG E P, et al.Prediction of troponin-T degradation using color image texture features in 10d aged beef longissimus steaks[J].Meat Science, 2014, 96(2):837-842
[19]陈坤杰, 季方芳, 徐佳琪.生理成熟度及牛肉肌纤维特征与嫩度关系试验研究[J].农业机械学报, 2018, 49(05):375-381
[20]王小川 .MATLAB神经网络43个案例分析[M]. 北京：北京航空航天大学出版社, 2013：120-150.
[21]于梦馨, 刘波, 汤恩生.改进粒子群算法优化参数的遥感图像分类[J].航天返回与遥感, 2018, 39(02):133-140
[22]王睿.关于支持向量机参数选择方法分析[J].[J].重庆师范大学学报（自然科学版）, 2007, 24(2):36-38
[23]STüN B, MELSSEN W J, OUDENHUIJZEN M, et al.Determination of optimal support vector regression parameters by genetic algorithms and simplex optimizationAnal Chim Acta[J].Analytica Chimica Acta, 2005, 544(1):292-305
[24]丁世飞, 齐丙娟, 谭红艳.支持向量机理论与算法研究综述[J].电子科技大学学报, 2011, 40(1):2-10
[25]PAL M, DESWAL S.Support vector regression based shear strength modelling of deep beams[J].Computers & Structures, 2011, 89(13):1430-1439
[26]YUVARAJ P, MURTHY A R, IYER N R, et al.Support vector regression based models to predict fracture characteristics of high strength and ultra high strength concrete beams[J].Engineering Fracture Mechanics, 2013, 98(1):29-43
[27]王健峰, 张磊, 陈国兴, 等.基于改进的网格搜索法的参数优化[J].应用科技, 2012, 39(3):28-31
[28]王卫.基于图像纹理和大理石花纹特征的牛肉嫩度智能预测方法研究[D]. 南京：南京农业大学, 2011.
[29]陈士进, 丁冬, 李泊, 等.基于机器视觉的牛肉结缔组织特征和嫩度关系研究[J].南京农业大学学报, 2016, 39(05):865-871
[30]王健峰.基于改进网格搜索法SVM参数优化的说话人识别研究[D].哈尔滨：哈尔滨工程大学, 2012.
[31]魏康丽, 王振杰, 孙柯, 等.基于计算机视觉对苹果脆片外观品质分级[J].南京农业大学学报, 2017, 40(03):547-555
[32]朱兴统, 左敬龙, 张晶华.改进量子粒子群优化支持向量机的网页分类[J].湖南科技大学学报自然科学版, 2012, 27(03):81-85