基于RGB颜色空间的冷冻猪肉储藏时间机器视觉判定
|
摘要
为解决冷冻猪肉储藏时间在人工判定中准确率和效率低的问题,该研究基于机器视觉和图像处理技术分析了冷冻猪肉表面图像RGB颜色特征,并将无光泽肌肉像素点比例作为自动判定冷冻猪肉储藏时间的特征参数。结果显示:将红度(R)与蓝度(B)的差值小于等于30且红度(R)与绿度(G)的差值小于等于30作为第一特征条件,将红度(R)与蓝度(B)的差值小于等于30、红度(R)与绿度(G)的差值小于等于30且蓝度(B)大于等于100作为第二特征条件对冷冻猪肉颜色特征进行提取时,二值化图像能较好地识别冷冻猪肉表面总肌肉和无光泽肌肉。基于统计学分析,当储藏时间在3个月以内(1、2和3个月)和超过12个月(13、15和17个月)的冷冻猪肉无光泽肌肉像素点比例阈值为26.8%时,冷冻猪肉验证检测判定准确度最佳,分别为90.00%和81.67%。经图像分割后提出的基于无光泽肌肉像素点比例的判定方法可以为实际应用提供技术支持和参考。
The storage time of chilled pork is one of the most important indexes to price of pork.However,storage time of chilled pork was determined by trainees with their own visual senses in most of the Chinese enterprises.This manual grading method demands not only intensive labor but also lacks objectivity and accuracy.The objective of this research was to develop an optimal method for determining the storage time of chilled pork based on computer vision and image processing technologies to meet the requirement of the meat industry.A practical algorithm that can be used in a chilled pork storage time determining system was proposed in this research.The chilled pork sample images were collected by a machine vision image acquisition system.The system consisted of an image acquisition device,computer,and image processing algorithm equipped into the self developed system software.The images of the chilled pork samples in a load table were captured by CCD.Light intensity was regulated through a light controller,and the distance between the camera lens and the chilled pork samples was adjusted though translation stages in the load table.Collected images were automatically stored in the computer for further image processing.First,some methods such as background removal and image denoising were adopted to preprocess the image to obtain a region of interest.In this step,the image was cropped to separate the chilled pork from the background.Then,a color difference method was used to segment the chilled pork area to obtain the area of total muscle and un-glossed muscle.Our results showed that the total and un-glossed muscle of chilled pork can be well distinguished by using binary images when R-B≤30,R-G≤30(first feature) and R-B≤30,R-G≤30,100≤B(second feature),respectively.Based on the first and second feature,the number of pixel in the binary images for the total and un-glossed muscle can be obtained from the color information in each pixel,and the ratio of pixel amount between un-glossed muscle and total muscle can also be obtained.We calculated the ratio of un-glossed muscle of different batches,screened the maximum ratio of un-glossed muscle of chilled pork that was stored less than three months and the minimum ratio of un-glossed muscle of chilled pork that was stored more than 12 months,respectively,and then calculated the mean of maximum and minimum value mentioned above.The mean was taken as the threshold value,and that was increased or decreased successively if the actual value was less than or equal to the threshold value mentioned above.The storage time was determined to be within three months if the actual value was greater than the threshold value,and the storage time was determined to be over 12 months.Finally,60 chilled pork that were stored less than three months and 60 chilled pork that were stored more than 12 months were used to verify the threshold value.Our results showed that,the accuracy of the verification and determination of chilled pork that were stored less than three months(90.00%) and chilled pork that were stored more than 12 months(81.67%) were considerably improved when the threshold value was 26.8%.The results showed that the algorithm can meet the requirements of frozen pork storage time.
The storage time of chilled pork is one of the most important indexes to price of pork.However,storage time of chilled pork was determined by trainees with their own visual senses in most of the Chinese enterprises.This manual grading method demands not only intensive labor but also lacks objectivity and accuracy.The objective of this research was to develop an optimal method for determining the storage time of chilled pork based on computer vision and image processing technologies to meet the requirement of the meat industry.A practical algorithm that can be used in a chilled pork storage time determining system was proposed in this research.The chilled pork sample images were collected by a machine vision image acquisition system.The system consisted of an image acquisition device,computer,and image processing algorithm equipped into the self developed system software.The images of the chilled pork samples in a load table were captured by CCD.Light intensity was regulated through a light controller,and the distance between the camera lens and the chilled pork samples was adjusted though translation stages in the load table.Collected images were automatically stored in the computer for further image processing.First,some methods such as background removal and image denoising were adopted to preprocess the image to obtain a region of interest.In this step,the image was cropped to separate the chilled pork from the background.Then,a color difference method was used to segment the chilled pork area to obtain the area of total muscle and un-glossed muscle.Our results showed that the total and un-glossed muscle of chilled pork can be well distinguished by using binary images when R-B≤30,R-G≤30(first feature) and R-B≤30,R-G≤30,100≤B(second feature),respectively.Based on the first and second feature,the number of pixel in the binary images for the total and un-glossed muscle can be obtained from the color information in each pixel,and the ratio of pixel amount between un-glossed muscle and total muscle can also be obtained.We calculated the ratio of un-glossed muscle of different batches,screened the maximum ratio of un-glossed muscle of chilled pork that was stored less than three months and the minimum ratio of un-glossed muscle of chilled pork that was stored more than 12 months,respectively,and then calculated the mean of maximum and minimum value mentioned above.The mean was taken as the threshold value,and that was increased or decreased successively if the actual value was less than or equal to the threshold value mentioned above.The storage time was determined to be within three months if the actual value was greater than the threshold value,and the storage time was determined to be over 12 months.Finally,60 chilled pork that were stored less than three months and 60 chilled pork that were stored more than 12 months were used to verify the threshold value.Our results showed that,the accuracy of the verification and determination of chilled pork that were stored less than three months(90.00%) and chilled pork that were stored more than 12 months(81.67%) were considerably improved when the threshold value was 26.8%.The results showed that the algorithm can meet the requirements of frozen pork storage time.
引文
[1]褚衍章,朱增勇.中国猪肉消费“十二五”回顾及“十三五”展望[J].农业展望,2016(5):81-85.
[2]杨益,何颖.基于RGB空间的车道线检测与辨识方法[J].计算机与现代化,2014(2):86-90.
[3]孙红,郭凯.融合字符纹理特征与RGB颜色特征的车牌定[J].光电工程,2015,42(6):14-19,44.Sun Hong,Guo Kai.License plate location based on characters texture feature and RGB color feature[J].Opto-Electronic Engineering,2015,42(6):14-19,44.(in Chinese with English abstract)
[4]郝永杰,周博文.基于图像特征及改进支持向量机算法的交通标志识别[J].现代电子科技,2017,40(8):97-99.Hao Yongjie,Zhou Bowen.Traffic sign recognition based on image feature and improved support vector machine algorithm[J].Modern Electronics Technique,2017,40(8):97-99.(in Chinese with English abstract)
[5]贾永红,谭慧,黄若冰.基于车载视频的道路交通指示标志的检测与识别[J].测绘地理信息,2016,41(5):47-50.Jia Yonghong,Tan Hui,Huang Ruobing.A detection and recognition on method for indicatory traffic signs based on vehicle videos[J].Journal of Geomatics,2016,41(5):47-50.(in Chinese with English abstract)
[6]冯范.基于颜色空间的鲜切果蔬图像褐变特征分析[D].信阳:信阳师范学院,2015.
[7]尹建军,王新忠,毛罕平,等.RGB与HSI颜色空间下番茄图像分割的对比研究[J].农机化研究,2006(11):171-174.
[8]郭俊先,李俊伟,胡光辉,等.新疆冰糖心红富士苹果RGB图像多指标分析[J].新疆农业科学,2013,50(3):509-517.Guo Junxian,Li Junwei,Hu Guanghui,et al.Analysis of multi-objective of Xinjiang Fuji apple by RGB image[J].Xinjiang Agricultural Sciences,2013,50(3):509-517.(in Chinese with English abstract)
[9]Chandel A K,Khot L R,Osroosh Y,et al.Thermal-RGBimager derived in-field apple surface temperature estimates for sunburn management[J].Agricultural and Forest Meteorology,2018,253-254(1):132-142.
[10]陈毅,刘晓玉,蒋峥,等.基于RGB和HSV的胶囊异囊缺陷识别方法[J].计算机工程与设计,2014,35(11):3888-3892.Chen Yi,Liu Xiaoyu,Jiang Zheng,et al.Capsule Yinang defect recognition based on RGB and HSV color space[J].Computer Engineering and Design,2014,35(11):3888-3892.(in Chinese with English abstract)
[11]Pfitzner C,May S,Nüuchter A.Evaluation of features from RGB-D data for human body weight estimation[J].International Federation of Automatic Control,2017,50(1):10148-10153.
[12]李凯旋,刘哲星,刘思德,等.胶囊内窥图像识别系统算法设计及实现[J].南方医科大学学报,2012,32(7):948-951.Li Kaixuan,Liu Zhexing,Liu Side,et al.Design and realization of an algorithm for capsule endoscope image recognition system[J].Journal of Southern Medical University,2012,32(7):948-951.(in Chinese with English abstract)
[13]徐婷婷,吉晓东,李文华,等.基于颜色和纹理特征的胶囊内镜图像分类[J].现代电子技术,2018,41(19):58-62.Xu Tingting,Ji Xiaodong,Li Wenhua,et al.Capsule endoscope image classification based on color and texture features[J].Modern Electronics Technique,2018,41(19):58-62.(in Chinese with English abstract)
[14]张洪铭,陈先锋,张英,等.基于RGB颜色模型的玉米淀粉爆燃火焰传播速度[J].爆炸与冲击,2018,38(1):133-139.Zhang Hongming,Chen Xianfeng,Zhang Ying,et al.Flame propagation velocities of cornstarch dust explosion based on RGB color model[J].Explosion and Shock Waves,2018,38(1):133-139.(in Chinese with English abstract)
[15]Belati A,Cajaiba J.Measurement of wax appearance temperature using RGB image analysis and FBRM[J].Fuel,2018,220(20):264-269.
[16]李倩.基于计算机视觉的船舶类型识别与流量监控算法研究[J].舰船科学技术,2018,40(10):151-153.Li Qian.Research on ship type recognition and flow monitoring algorithm based on computer vision[J].Ship Science and Technology,2018,40(10):151-153.(in Chinese with English abstract)
[17]王磊.计算机视觉在船舶焊缝缺陷识别的应用[J].舰船科学技术,2018,40(8):196-198.Wang Lei.Application of computer vision in identification of ship weld defects[J].Ship Science and Technology,2018,40(8):196-198.(in Chinese with English abstract)
[18]潘婧,钱建平,刘寿春,等.计算机视觉用于猪肉新鲜度检测的颜色特征优化选取[J].食品与发酵工业,2016,42(6):153-158.Pan Jing,Qian Jianping,Liu Shouchun,et al.Optimized the color parameters in prediction of pork freshness by computer vision[J].Food and Fermentation Industries,2016,42(6):153-158.(in Chinese with English abstract)
[19]赵杰文,方明,刘木华,等.基于Ohta和RGB颜色空间牛胴体背长肌的分割[J].江苏大学学报(自然科学版),2006,27(3):189-192.Zhao Jiewen,Fang Ming,Liu Muhua,et al.Segmentation of longissmi us dorsi in beef carcass based on Ohta and RGBcolor systems[J].Journal of Jiangsu University(NaturalScience Edition),2006,27(3):189-192.(in Chinese with English abstract)
[20]Orava J,Parkkinen J,Markku H K,et al.Temporal clustering of minced meat by RGB and spectral imaging[J].Journal of Food Engineering,2012,112(2):112-116.
[21]张雷蕾,李永玉,彭彦昆,等.基于高光谱成像技术的猪肉新鲜度评价[J].农业工程学报,2012,28(7):254-259.Zhang Leilei,Li Yongyu,Peng Yankun,et al.Determination of pork freshness attributes by hyperspectral imaging technique[J].Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE),2012,28(7):254-259.(in Chinese with English abstract)
[22]张萌,钟南,刘莹莹.基于生猪外形特征图像的瘦肉率估测方法[J].农业工程学报,2017,33(12):308-314.Zhang Meng,Zhong Nan,Liu Yingying.Estimation method of pig lean meat percentage based on image of pig shape characteristics[J].Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE),2017,33(12):308-314.(in Chinese with English abstract)
[23]周炜.基于多源信息融合技术的猪肉新鲜度无损检测方法研究[D].武汉:华中农业大学,2009.Zhou Wei.Based on the Multi-source Information Fusion Technology of Pork Freshness NDT Methods Research[D].Wuhan:Huazhong Agricultural University,2009.(in Chinese with English abstract)
[24]Dowlati M,Mohtasebi S S,Omid M,et al.Freshness assessment of gilthead sea bream(Sparus aurata)by machine vision based on gill and eye color changes[J].J Food Engineer,2013,119(2):277-287.
[25]王红梅,冯义东.基于改进活动轮廓模型的图像边界分割[J].内蒙古师范大学学报,2016,45(1):17-20,25.Wang Hongmei,Feng Yidong.Image edge segmentation based on improved active contour model[J].Journal of Inner Mongolia Normal University(Natural Science Edition),2016,45(1):17-20,25.(in Chinese with English abstract)
[26]吴国伟,谢金法,郭志强.基于Sobel算子的车辆轮廓边缘检测算法[J].河南科技大学学报,2009,30(6):39-41.
[27]杨成佳.图像去噪及其效果评估若干问题研究[D].长春:吉林大学,2016.Yang Chengjia.Study on Image Denoising and Its Effect Evaluation[D].Changchun:Jilin University,2016.
[28]王丹竹,田科雄.冷冻肉品质与时间和温度变化的关系[J].中国畜禽种业,2012(8):24-26.
[29]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.Chen Kunjie,Yin Wenqing.Application of machine vision technology in quantitative analysis of color features of beef lean tissue[J].Food Science,2008,29(9):92-96.(in Chinese with English abstract)
[30]黄新波,张慧莹,张烨,等.基于改进色差法的复合绝缘子图像分割技术[J].高电压技术,2018,44(8):2493-2500.Huang Xinbo,Zhang Huiying,Zhang Ye,et al.Composite insulator image sementation technology based on improve color difference[J].High Voltage Engineering,2018,44(8):2493-2500.(in Chinese with English abstract)
[31]王辉,田寒友,李文采,等.基于顶空气相色谱-离子迁移谱技术的冷冻猪肉储藏时间快速判别方法[J].食品科学,http://kns.cnki.net/kcms/detail/11.2206.TS.20180821.1418.072.html Wang Hui,Tian Hanyou,Li Wencai,et al.Fast discriminating method of storage time of frozen pork based on HS-GC-IMS[J].Food Science,http://kns.cnki.net/kcms/detail/11.2206.TS.20180821.1418.072.html
[2]杨益,何颖.基于RGB空间的车道线检测与辨识方法[J].计算机与现代化,2014(2):86-90.
[3]孙红,郭凯.融合字符纹理特征与RGB颜色特征的车牌定[J].光电工程,2015,42(6):14-19,44.Sun Hong,Guo Kai.License plate location based on characters texture feature and RGB color feature[J].Opto-Electronic Engineering,2015,42(6):14-19,44.(in Chinese with English abstract)
[4]郝永杰,周博文.基于图像特征及改进支持向量机算法的交通标志识别[J].现代电子科技,2017,40(8):97-99.Hao Yongjie,Zhou Bowen.Traffic sign recognition based on image feature and improved support vector machine algorithm[J].Modern Electronics Technique,2017,40(8):97-99.(in Chinese with English abstract)
[5]贾永红,谭慧,黄若冰.基于车载视频的道路交通指示标志的检测与识别[J].测绘地理信息,2016,41(5):47-50.Jia Yonghong,Tan Hui,Huang Ruobing.A detection and recognition on method for indicatory traffic signs based on vehicle videos[J].Journal of Geomatics,2016,41(5):47-50.(in Chinese with English abstract)
[6]冯范.基于颜色空间的鲜切果蔬图像褐变特征分析[D].信阳:信阳师范学院,2015.
[7]尹建军,王新忠,毛罕平,等.RGB与HSI颜色空间下番茄图像分割的对比研究[J].农机化研究,2006(11):171-174.
[8]郭俊先,李俊伟,胡光辉,等.新疆冰糖心红富士苹果RGB图像多指标分析[J].新疆农业科学,2013,50(3):509-517.Guo Junxian,Li Junwei,Hu Guanghui,et al.Analysis of multi-objective of Xinjiang Fuji apple by RGB image[J].Xinjiang Agricultural Sciences,2013,50(3):509-517.(in Chinese with English abstract)
[9]Chandel A K,Khot L R,Osroosh Y,et al.Thermal-RGBimager derived in-field apple surface temperature estimates for sunburn management[J].Agricultural and Forest Meteorology,2018,253-254(1):132-142.
[10]陈毅,刘晓玉,蒋峥,等.基于RGB和HSV的胶囊异囊缺陷识别方法[J].计算机工程与设计,2014,35(11):3888-3892.Chen Yi,Liu Xiaoyu,Jiang Zheng,et al.Capsule Yinang defect recognition based on RGB and HSV color space[J].Computer Engineering and Design,2014,35(11):3888-3892.(in Chinese with English abstract)
[11]Pfitzner C,May S,Nüuchter A.Evaluation of features from RGB-D data for human body weight estimation[J].International Federation of Automatic Control,2017,50(1):10148-10153.
[12]李凯旋,刘哲星,刘思德,等.胶囊内窥图像识别系统算法设计及实现[J].南方医科大学学报,2012,32(7):948-951.Li Kaixuan,Liu Zhexing,Liu Side,et al.Design and realization of an algorithm for capsule endoscope image recognition system[J].Journal of Southern Medical University,2012,32(7):948-951.(in Chinese with English abstract)
[13]徐婷婷,吉晓东,李文华,等.基于颜色和纹理特征的胶囊内镜图像分类[J].现代电子技术,2018,41(19):58-62.Xu Tingting,Ji Xiaodong,Li Wenhua,et al.Capsule endoscope image classification based on color and texture features[J].Modern Electronics Technique,2018,41(19):58-62.(in Chinese with English abstract)
[14]张洪铭,陈先锋,张英,等.基于RGB颜色模型的玉米淀粉爆燃火焰传播速度[J].爆炸与冲击,2018,38(1):133-139.Zhang Hongming,Chen Xianfeng,Zhang Ying,et al.Flame propagation velocities of cornstarch dust explosion based on RGB color model[J].Explosion and Shock Waves,2018,38(1):133-139.(in Chinese with English abstract)
[15]Belati A,Cajaiba J.Measurement of wax appearance temperature using RGB image analysis and FBRM[J].Fuel,2018,220(20):264-269.
[16]李倩.基于计算机视觉的船舶类型识别与流量监控算法研究[J].舰船科学技术,2018,40(10):151-153.Li Qian.Research on ship type recognition and flow monitoring algorithm based on computer vision[J].Ship Science and Technology,2018,40(10):151-153.(in Chinese with English abstract)
[17]王磊.计算机视觉在船舶焊缝缺陷识别的应用[J].舰船科学技术,2018,40(8):196-198.Wang Lei.Application of computer vision in identification of ship weld defects[J].Ship Science and Technology,2018,40(8):196-198.(in Chinese with English abstract)
[18]潘婧,钱建平,刘寿春,等.计算机视觉用于猪肉新鲜度检测的颜色特征优化选取[J].食品与发酵工业,2016,42(6):153-158.Pan Jing,Qian Jianping,Liu Shouchun,et al.Optimized the color parameters in prediction of pork freshness by computer vision[J].Food and Fermentation Industries,2016,42(6):153-158.(in Chinese with English abstract)
[19]赵杰文,方明,刘木华,等.基于Ohta和RGB颜色空间牛胴体背长肌的分割[J].江苏大学学报(自然科学版),2006,27(3):189-192.Zhao Jiewen,Fang Ming,Liu Muhua,et al.Segmentation of longissmi us dorsi in beef carcass based on Ohta and RGBcolor systems[J].Journal of Jiangsu University(NaturalScience Edition),2006,27(3):189-192.(in Chinese with English abstract)
[20]Orava J,Parkkinen J,Markku H K,et al.Temporal clustering of minced meat by RGB and spectral imaging[J].Journal of Food Engineering,2012,112(2):112-116.
[21]张雷蕾,李永玉,彭彦昆,等.基于高光谱成像技术的猪肉新鲜度评价[J].农业工程学报,2012,28(7):254-259.Zhang Leilei,Li Yongyu,Peng Yankun,et al.Determination of pork freshness attributes by hyperspectral imaging technique[J].Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE),2012,28(7):254-259.(in Chinese with English abstract)
[22]张萌,钟南,刘莹莹.基于生猪外形特征图像的瘦肉率估测方法[J].农业工程学报,2017,33(12):308-314.Zhang Meng,Zhong Nan,Liu Yingying.Estimation method of pig lean meat percentage based on image of pig shape characteristics[J].Transactions of the Chinese Society of Agricultural Engineering(Transactions of the CSAE),2017,33(12):308-314.(in Chinese with English abstract)
[23]周炜.基于多源信息融合技术的猪肉新鲜度无损检测方法研究[D].武汉:华中农业大学,2009.Zhou Wei.Based on the Multi-source Information Fusion Technology of Pork Freshness NDT Methods Research[D].Wuhan:Huazhong Agricultural University,2009.(in Chinese with English abstract)
[24]Dowlati M,Mohtasebi S S,Omid M,et al.Freshness assessment of gilthead sea bream(Sparus aurata)by machine vision based on gill and eye color changes[J].J Food Engineer,2013,119(2):277-287.
[25]王红梅,冯义东.基于改进活动轮廓模型的图像边界分割[J].内蒙古师范大学学报,2016,45(1):17-20,25.Wang Hongmei,Feng Yidong.Image edge segmentation based on improved active contour model[J].Journal of Inner Mongolia Normal University(Natural Science Edition),2016,45(1):17-20,25.(in Chinese with English abstract)
[26]吴国伟,谢金法,郭志强.基于Sobel算子的车辆轮廓边缘检测算法[J].河南科技大学学报,2009,30(6):39-41.
[27]杨成佳.图像去噪及其效果评估若干问题研究[D].长春:吉林大学,2016.Yang Chengjia.Study on Image Denoising and Its Effect Evaluation[D].Changchun:Jilin University,2016.
[28]王丹竹,田科雄.冷冻肉品质与时间和温度变化的关系[J].中国畜禽种业,2012(8):24-26.
[29]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.Chen Kunjie,Yin Wenqing.Application of machine vision technology in quantitative analysis of color features of beef lean tissue[J].Food Science,2008,29(9):92-96.(in Chinese with English abstract)
[30]黄新波,张慧莹,张烨,等.基于改进色差法的复合绝缘子图像分割技术[J].高电压技术,2018,44(8):2493-2500.Huang Xinbo,Zhang Huiying,Zhang Ye,et al.Composite insulator image sementation technology based on improve color difference[J].High Voltage Engineering,2018,44(8):2493-2500.(in Chinese with English abstract)
[31]王辉,田寒友,李文采,等.基于顶空气相色谱-离子迁移谱技术的冷冻猪肉储藏时间快速判别方法[J].食品科学,http://kns.cnki.net/kcms/detail/11.2206.TS.20180821.1418.072.html Wang Hui,Tian Hanyou,Li Wencai,et al.Fast discriminating method of storage time of frozen pork based on HS-GC-IMS[J].Food Science,http://kns.cnki.net/kcms/detail/11.2206.TS.20180821.1418.072.html