计算机视觉技术在农业上的应用
2024-08-02
来源:步旅网
Agricultural Science &Technology,201 7,1 8(1 1):21 58—21 62 Copyright@201 7,Information Institute of HAAS.All rights reserved Agricultural Information Science Application of Computer Vision Technology in Agriculture Ximei HUANG’,Jianjie BI ,Nan ZHANG’,Xiaoling DING卜,Fei LI’,Fadong HOU’ 1.College of Mechanical and Electronic Engineering,Shandong Agricultural University,Shandong Provincial Key Laboratory of Horticultural Machineries and Equipments,Tai’an 271 01 8,China; 2.ChinaState Key Laboratory of Crop・Biology,Shangdong Agricultural Universiy,Tai’an City 271t 01 8,China Abstract With the development of image processing technology and computer, computer vision technology has been widely used in the production of agriculture, and has made many jmpo ̄ant achievements.This paper reviews its research progress on diagnosis of agricultural products,water diagnosis,weed identification, 计算机视觉技术在农业上的应 用 黄喜梅1 9毕建杰2张楠 ,丁筱玲 ,李飞 ,侯发 东。 (1.山东农业大学机械与电子工程学 院,山东省园艺机械与装备重点实验室,泰 重点实验室.山东泰安27lOl8) product quaily ttesting and grading.agricultural picking and sorting and other as- pects.and finally put forward its existing problems and prospects for lhe future. Prospect Key words Image processing;Computer vision technology;Agriculture production; 安271018:2.山东农业大学作物生物学国家 摘要随着图像处理技术和计算机的发 展。计算机视觉技术在农业的生产中有了更 广泛的应用。并且均取得了许多重要成果. C osthcmiepnucter svtiusdioynin tge chhonwo ltog my iask ae p aincdk ibnrgi eaflnyd p srobrteindg i natno dit so tdhevre alospmecetsnt, e machine“see”.In other prospects. words.It is the machine vision which 综述了农产品缺素诊断、水分诊断、杂草识 别、产品品质检测和分级、农业采摘与分选 等方面的研究进展.最后提出了其存在的问 题和对未来的展望。 uses computer and video camera to Application of Computer 关键词 图像处理;计算机视觉技术:农业生 产:前景 replace human eyes to identify,judge, measure and track the targets,and then proceed image processmg by us。 Vision Technology in Agri・ CuIture Research on computer vision tech- mg computer to process the tmages into those suitable for human observa— tion or to transmit for Instrument de. tection.Wilh the function of computer nology in nutrient deficiency diag- nosis The nutrients needed by crops are the food of their lives.which there. fore play a crucial role in their physic— logical status.If the nutrient elements of a crop are deficiency,then the growth process of the crop will be vision technology becoming more and more powerful,it has been more and more widely applied in the various fields in daily life.which reduces the human labor intensity.improves the productivity of mankind,realizes the intelligent and automatic agricultural production.The research on computer vision technology started Iate in China. and the technology is not advanced enough,neither do the research com. plete enough.However,a certain achievements have been made with the continuous progress of the various greatly affected,resulting in problems like fruitlessness。yellow leaves,short plants.Therefore.it is very necessaw to study the nutrient deficiency of crops. CAO{’1 used image processing technology to extract the color and texture CharaCteristlcs of soybean leaves,obtaining 9 groups of color characteristic parameters and 5 technologies in China.In order to pro. mote the jncrease of agriculturaI groups of texture parameters.and then used principal component analy- sis and multivariate linear regression method to establish soybean plant ni. trogen discriminant mode1.The verifi. cation results show that the modeI had 作者简介黄喜梅(1991一).女,山东菏泽人 研究生,研究方向为机器视觉.E—Mail xmhuang36@126.corn。 通讯作者 mechanization Ievelin China.this pa. per reviewed its research progress on agriculturaI products diagnosis,water diagnosis,weed identification,product quality testing and grading,agricultural 收稿日期修回日期2017—08—2l 2017—10—16 2017■ 2159 a certain precision and could be used for the rapid detection and analysis of nitrogen in soybean.In the research of ments in tomato Ieaves according to the color。texture and morphologicaI the color features of tomato Ieaves. Research on computer vision tech- characteristics of wheat through cap- turing the images of individua1 wheat diagnosing tomato disease of nutrient nology in moisture diagnosis beings.but also vitaI for crops.Nowa— kernels and bulk wheat samples of dif- ned the collected feature parameters deficiency intellectively in the soilless moisture contents,and scree- Water is not only vitaI for human ferentagriculture.XU【。】used genetic algo- rithm to select features of Ieaves im・ age in order to get best information for diagnosing,which achieved the de— days,water deficit is a common prob— Iem In the process of agricultural pro- duction,and water deficit has a great using stepwise discriminant analysis, finding that the overalI identification accuracy of BP neuraI network modeI reached 92%to the bulk wheat im- binationaI algorithm of homomorphic sired effect.ZHANG et a/.嘲analyzed effect on the plant。which can result In relative ratios rgb。as welI as the cor- quality reduction.However,too much turaI product yield decrease and ages.JIANG et aL[坷adopted the com- the red。green,blue(RGB)and their agdculrelations among nitrogen,phosphorus water wilI Iead to waste of water re. filtering and multi—scale Retinex for.I_ and water content of the Ieaves and their color parameters,and the results showed that there were high Iinear correlation between the nitrogen con- tent and the green weight G and Ieat chroma H.2007.LI【4】ext卜acted the RGB and the combined color charac- te rislic values of tomato images using image processing software,which were conducted with regression anal- ysis with the real-time nitrogen indica- tOrs.and the estimation model for n.- trogen content in diferent Ieaves were established by combining percent ground cover of vegetation(PGCV) with nitrogen nutrition parameters. WU is]used RGB color modeI to dis- criminate walnut deficiency with chro- ma as key features with the foIlowing process:first.convert Ieaves image to grayscale,and calculate threshold by using the Ieast squares method and segment image with this threshold to make the distinction between normal and deficiency regions,at last,com- pare pre-pixel gray in blade area with the typical gray values deriving fr0m statisticaI analysis to determine the missing elements of walnut.XU哟 proposed the nutrient deficiency intel- Iigent diagnose based on the color featUreS of rapeseed Ieaves through analyzing and extracting the changes of color,shape,texture and other ex- temaI features at diferent pads of rapeseed plants,and BayeS classifier was used to train and classify the fea- tures,finding that the nutrient deficien- cy diagnostic rate reached as high as 87.5%.and the diagnostic accuracy of phosphorus deficiency reached 1 oo%. TlAN ef aL【7l extracted the color image features of tomato leaves through the extemaI characteristics of Ieaf struc- ture.Ieaf color and leaf vein distribu- tion。and identified the missing ele- sources.Therefore.it Is urgent to im- prove the utilization ratio of water re— sources and make the crops have no water deficit. SUN et a^嗍established the image collection and monitodng system of water content in cucumber Ieaves us— ing Photoshop and MATLAB.deter- mined the cucumber Ieaf image fea— tures under diferent light sources through experiment,mapped the rela- tion curve between Ieaf water content and=mage gray scale gradient under conditions of most suitable back- ground light,established the prediction modeI for water content in cucumber Ieaves。and determined the irrigation regions through criticaI point expen. ment。thus reaching the non-destruc— tive testing of Ieaf water shortage by image character of cucumber leaves. GAN et a 嘲performed the image seg— mentation algorithm to images of fresh tobacco Ieaves with Matlab software。 extracted the parameters of r,g,H。 acreage,width,shrinkage rate,mean value,consistency and entropy which were used as the indexes of compre- hensive evaluation and ranked with the three—demarcation AHP method, finding that Ieaf width.Ieaf area and mean value had relatively great weight and,therefore。were recommended as the indicators in rapid determination of waler content of flu-cured tobacco Ieaves.DUAN et aL【fq adopted image processing technology establishing the BP neural network model and the ge— netic algodthm based Ieast squares supporf vector machine modeI accord- ing to the color characteristics and vein characteristics of tobacco Ieaves at diferent stages during bulk curing pro— cess and fresh Ieaves。finding that the models could estimate the water con— tent accurately.ZHANG[11】extracted lumination enhancement processing to extract image features of color。texture and morphology,correlation analysis and hypothesis testing were used for remarkable features selection。and fi— nally partiaI least squares regression was adopted to establish the water content detection mode1.ZHOU【13】 processed the corn grain images with average filtering。 maximum between-class variance(Otus)method and morphologicaI operation method,, and used MATLAB software program- ming to extract the external character- istics of morphology,texture and color of corn grains。and finally identified 8 principal components factors using stepwise discrlminant anaIysjs and p rincipal cOmpOnent anaIysis,thereby estabI.shing a three-IaVer BP neuraI netw0rk m0deI f0r the Dredicali0n 0f water c0ntents in c0m.GENG ef Dr0p0sed a cOttOn water deficit rea卜 time mOn.torina and aut0matic water and nutriti0n suDDIV sVstem based 0n cOmputer VisiOn techn010gy. ReSearCh On COmpUter vision tech- noIOgy in w咧r cogn.tion ln agriCUlturaI pnDduCti0n,weedS seriOusIy affect the ar0、『I,th 0f cr0ps。 and must be c0ntrO¨ed in a timeIV and e仟eCtive manner.Alth0uah pesticIde weedjng has high efflciency。 It has caused a great dea1 0f pestIcide waSte and caused Se ri0us harm tO the ecO- 10gicaI envir0nment.Theref0re, .t is very impOrtant t0 find a bener fIeld weed ident_fIcati0n techn0l0gy. ’ MAO ef a 【1司deve10ped a sOft- ware system tO transform c0l0r image to bina imageS by the COIOr feature 0f pIants and backgnDund, d t0 dIstin- guish c Ops and inte卜 Dw weeds by the locati0n fealure 0f cr0D w.th-n the fiIed.The results shOwed that the cOr— rect cIassmcatiOn rate of the sVstem 2017 was 86%.MAO et aL【1q transformed the color jndex and threshold for weed living standard,the quality of agncul・ tural products has higher requie- rkey Iink to ensure the quality.The sur- wilI have a serious impact on the qual- image enhancement,image prepro- cessing methods。such as threshold medin fialtering method of noise re- tion based on histogram。Laplarice al— image segmentation Into the segmen。 tation surface in RGB color space. used the evaluating method of seg- mentation error based on Bayes for- ments.and the quality inspection is the segmentation,edge extraction,chose face damage of agricultural prducots duction,overalI threshold segmenta- ityof products,thus affecting people’s use.1t iS of great signifiance to per- cforrn nondestructive testing and clas- mula to optimize the color indaxes and threshold parameter.thereby improv— gOrithm f0r edge detection,developed the binary regression modeI between the soluble solids ontentc of watermel- ing the segmentation accuracy of weed image.Burks et a1.n7l conducted accurate identification of the weeds in sunflower fields using BP neuraI net- work algorithm。and the identifica- sifiaticon of agdcuitural products.The on and density and mass of watermel- d experiments to following are some recent research on。and conducteresults. WANG【 】first captured the im- verify and analyze the accuracy of the method to estimate watermelon vol- tion rate of weed could reach up to ages of tomateos through computer 92%through the selection of proper vision system。which were processed parameters and network mode1.WU et denoising,segmentation and image 旧proposed a gray processing algo- enhancement。then extracted the fea- rithm based on the features of com- tures of tomatoes images.1ncluding 8 weed color jmages to denoise the gray feature parameters of 3 features.and image after transformed jnto graY lm- finally employed the improved BP ages,used Laplasse operator for edge neuraI algor|thm to train the tomatoes extraction, and finally used SVM with bruise for non,destructive classifi- method for classification。finding that the method had better identifiaction ef- cation.The experiments showed that the rate of testing precision and classi— fect than BP neuraI network with prop- fication was no Iess than 90%.WU【捌 er selected kemeI function,and the conducted the pre-processing analysis recognition rate reached as high as 98.3%.YANG ef aL l1碉pre・processed to rice images with image transform and image enhancement,then trans- the field weed using noise fitiedng and image enhancement.and then adopt- formed the rice images to HSl color ed artificiaI neuraI network algorithm space,getting their H-component his- for image paffem recognition,obtain・ togram,then used H-component his・ ing a good recognition effect.JIN et aLm togram,principal component analysis proposed a new segmentation algo- and genetic algorithm to extract the rithm of image to discdminate weeds riec image features.and finally classi- fr0m background based on color Infor- flde the rice images with Iinear dis- mation in RGB space,and applied 8- criminant function and BP neuraI net- neighborhcod method to eliminate the work.The method improved the clas- isolated points。thereby getting the Io- s讯cation accuracy.WANG【 pre・pro- cations of weed regions.The results cessed the captured corn images fisrt showed that the algorithm could iden— with gray processing,medina filtering, t;fy the weeds effectively.On this basis. image segmentation,morphological combined with automatic navigation te- processing,then extracted the profile, chnology.the structure modeI of weed・ geometriacI feature and color feature ing robot was designed.FENG ef a£ of com seed,and finally proposed a applied aimpace median filter to the multi-object features e)cIarction and image preprocesslng according to the optimized neural network using PCA i・ position feature of greenhouse plants, dentifiaction method adapting to com and then the combined use of colori- seeds varieties jdentification.The test metric method,maximum variance ersults showed that the integrated auto threshold method and seed filling identifiaction accuracy of the method algorithm were applied forimage bina- was more than 97%.×U ef .囝estab・ rization and separating the weeds.The Iished the fru|t shape classifiaction correct rceognition rate of weeds was system with the apples and the peare 89%.and the error recognition rate from Dangshan as examples through was3%. database construction。image pro- Research on computer vision tech- cessing, classifiaction, modeling, nology In product quality Inspec- matching and classification,and the tion and classification testing effect was good.ZHAO闽stu. With the improvement of people’s died watermelon image denoising, ume,and the accuracy of extracting and identifying deficient elements with the Identification accuracy reaching up tO 93.3%.HU唧obtained tea imagas using computer vision tcehnology, conducted a sedes processing to the tea images to extract the data param- eters of extemal characteristics,and ifnally usedBP neuraI networktoclas- sjyf the tea quality. Research on computer vision tech- nology in agricultural plcldng and sorting For thousands of years,fruit and vegetable picking is the most time- consuming part in agncuitural produc- tion。and in order to solve this problem。 the agdculturaI robot comes Int0 being. The agricuItural robot plays an irre- placeable roleintheprocesofpicking fruits and vegetables.which can im・ provetheproductivity,reducethecost of picking,ensure better picking quail— ty,s0 it plays a very important role in agdcuRural production.Therefore,jt seems to be very necessary to study agricuRuraI robot based on machine vision. The Institute of AgriculturaI and EnvironmentaI Engineenng of Nether- Iands ̄successfully developed a mo- bile cucumber harvesting robot proto- type,and}ts vision system∞uId de teGt the p ise pos.tjOn 0f mature fm-t' which had aChieved gOOd resuIts in greenhOuse and Iaborat0ry pic ng. LIU ef a I糊successfuIly deveIOped a fuIIy autOmatic gra协na svstem for cu- cu ItaceOus vegetabIes,and the s s- tem c0uId iudge the directiOn 0f seed¨ng a nd seed¨ng quaIity,which reallzed the aut0matic grafting of cucur- bit{lceous Veg Iables with good quaI ensur薷 ZHANG ef a 嗍adODted LR. CD meth0d t0 dissecl the image 0f strawber , and then extracted the Z1石1 features of strawberry image.so as to recognRion.but how to segment im— Sciences,2015。37(4):576—582. determine the barycenter of strawberry ages tr0m tast moving=mages and ex。 and the plucking position.ZHANG et tract effective features is still a problem a1.田1 developed a set of non-destruc- tive automatic picking executive body of strawberry,which could distinguish and Icaote dpe strawberries to a cer- tain range.and the fruits were har- vested by gnpping and curing pedun— cles sO as to achieve scatheless pick- ing.The results showed that the SUC- cess ratio of scatheless picking was that has not been solved yet. 【7】TIAN XL(田秀丽),HUANG YL(黄亚丽). sludV on the color image of tomato leaves under the computer vision sys- Computer vision technology has been more and more widely used in the process of agriculturaI prductioon。 promising with a broad prospect,but the diversity of the ecologicaI envion-r tem f计算机视觉系统下缺索番茄叶片彩 色图像研究)【J】.JoumaI of Agricultural Mechanization Research。201 7.39(7】: 175_.179. [8】SUN RD(孙瑞东),YU HY(于海业).Re- search on non-destructive detecting of ment makes its applications in agncul— ture more compliated than icn other cucumber Ieave water content based on areas,so it has higher requiements ron the accuracy of technologies.Im- proving the identification precision of image processing(基于图像处理的黄瓜 叶片含水量无损检测研究)【J】.Joumal fo Agricultural Mochanization Research。 2008. more than 90%.SUN et aL【弼devel・ oped a wheeled robot system for auto- matic detection and picking ripe ap— pies,which could realize automatic navigation。identification,picking func- tions in the case of unattended,and the results showed that the correct rate of apple rceogn ̄ion was as high as 94%and picking success rate reached 91.33%.TANG ef a1.阁pro- posed a kind of identification method 0f the tea ridge and navigation method of the tea-plunking machine,and the expenmentaI results showed that this method could solve the disadvantages of curing Ieaves with too many old Ieaves.LI刚used binocular stereo vi— sion system to identify red tomatoes, segmented the imagines by threshold segmentation method after noise re- moval using median filtering,gray。 scale image processing,and circular Hough transform algorithm was used to extract the center co0rdinates and radius characteristics of tomatoes, thereby achieving the extraction of fruit characteristics.The test results showed that the rceognition accuracy rate of ripe red tomatoes of the system could reach over 99%.which could fully meet the requierments of picking. Existing Problems in COm- puter Vision Technology and Its Prospects ln the rapid development process fo computer vision technology,there are many problems that must be solved:fimt,in many applications, some image algorithms。for example, image restoration,image segmenta‘ tion。feature extraction and paRern recognition,have a Iot of problems, which are in urgent need to find a bet・ ter algorithm to solye this problem. Second,most of the studies are based on static image models and paRem current=mage processing is the prob‘ lemthatisin urgentneedtobesolved. It is stilI the focus in future research of finding the high efficient algorithm to get images reseamh,as well as realiz- lng the combination with other soft- ware designs to make machine vision technology eb more widely applied. References 【1】CAO M(曹敏).Detecting research of soybean’s plant nitrogen based on im- age processingtechnology(基于图像处 理技术的大豆植株氮素测定研究)【D】. Jilin University.2005. [2】XUGL(徐贵力),CHENGYH(程月华), MAO HP(毛罕平).Features selection of leaves image for diagnosing tomato dis— ease of nutrient deficiency based on ge・ netic algo ̄hm(基于遗传算法的番茄缺 素叶片图像特征选择)[J】.Computer En- gineedng。2003,29(11):129—131. 【3】ZHANG YE(张彦娥),LI MZ(李民赞), ZHANG×J(张喜杰).Nutrition informa- tion extraction of the cucumber Ieaves in the greenhouse based on computer vi— sion technolgoY f基于计算机视觉技术的 温室黄瓜叶片营养信息检测)[J】.Trans- actions of the Chinese Society of Agd- culturaI Engineering,2005,21(8):102— 105. 【4】LIY(李艳),WANG J(王娟),MATF(马 腾飞).A primary study on N evaluating of processing tomato leaves based on color fe ̄uras(基于颜色特征的加工番 茄叶片氮素评价初步研究)【J1.Acta Agri— culturae BorealI—Occidentatis Sinica. 2007.16f3):175—17g. 【5】WUY(吴艳),HONGM(洪明),MENG XY(孟晓艳).Walnut nutrient deficiency discrimination method based on RGB oclor modeI(基于RGB颜色模型的核桃 缺素症判别方法)[J】.JoumaI fo Beihua University(NaturaI Science)2013(4): 493—d96. 【6】XU SY(徐胜勇),LIN WG(林卫国),WU WB(伍文兵).Nutdent deftciency image dlagnose of rapeseed based on color femurs f基于颜色特征的油菜缺素症图 像诊断)[J】.Chinese JoumaI of OiI Crop 【9】GAN LP(甘露萍),XIE SY(谢守勇), ZOU DJ(邹大军).Computer vision. based non-desturctive determination of watercontentinfreshtobaccoleavesof flue-cured tobacco and its implemenm- tion in MATLAB(基于计算机视觉的烤烟 鲜烟叶含水量无损检测及MATLAB实 现)【J】.JoumaI of Southwest Universiyt (Natural Science Edition)。2009.31(7): 166—170. 【10]DUAN SJ(段史江),SONG CP(宋朝 鹏),MAL(马力).Prdeitcionoftobacco Ieaf’s water contents during bulk cur- ing process based on image process- ing technique f基于图像处理的烘烤过 程中烟叶含水量检测)【J】.JoumaI of Northwest A&F Universiyt fNatural cSience Edition),2012(5): 4—80. 【1 1】ZHANG YR(张玉荣),CHEN SS(陈赛 赛),ZHOU XQ(周显青).Identification method of moisture content of wheat basedonimageprocessing(基于图像 处理的小麦水分含量识别方法研究)[J】. JoumaI of Henan Universiyt of Tech- nolgoy(NaturaI Science Edition)。2014。 35(5):101-106. 【12]JIANG CH(江朝晖),YANG CH(杨春 合),ZHOU Q(周琼).Detection of canopy water content of winter wheat during wintering perido based on image fe=ures(基于图像特征的越冬期冬小 麦冠层含水率检测)【J1.Transactions of the Chinese Society for Agricultural Machinery,201 5,46(1 2):2I -267. 【13]ZHOU HD(周鸿达),ZHANG YR(张玉 荣),WANG WY(王伟宇).Water con— tent deIection method of com grain based on image naalysis(基于图像处 理玉米水分检测方法研究)[J】.JoumaI of Henan Universiyt of Technology (Natural cSience Edition)。2016。37(3): 9B一100. 【14]GENG BT(耿宾涛),WANG XJ(王现 君1.Cotton automatic water nad nutd- tion supply system design based on computer vision(计算机视觉运用于棉 花水分养分供给系统设计)【J】.Joumal of AgriculturaI Mechanization Re- search.2017,39(1 1):179—182. 『1 51 MAO W。WANG Y。ZHANG X.Ma. chine vision system used for real-time detection inter-row weeds『J1.Transec- Z1 2O17 tions of the Chinese Society of Agricul- jdent计ication and classification of CA0 QX(曹其新).Study on robot vi— turaI Engineering,2o03。1 9(5):1 1 4—11 7. tomatoes with bruise using computer sion and iudgment system for seedling 【1 6]MAO HP(毛罕平)。HU B(胡波), vision(基于计算机视觉的番茄损伤自 grafting robot(秧苗嫁接机器人视觉与 ZHANG YC(张艳诚).Optimization of 动检测与分类研究)【J】.Transactions of 识别的研究)[J】.Transactions of the ooler index and threshold segmant ̄ion the Chinese Society fo AgriculturaI En・ Chinese Society for AgriculturaI Ma-  ̄VMred recognition(杂草识别中颜色特 gineedng,2005,21(8):98-1 01. chinery,2001。32(4):38—41. 、征和瞬值分割算法的优化)【J】.Transac- 【23】wu F(吴飞).Stduy on the application 【301 ZHANG TZ(张铁中),CHEN LB(陈利 of machine vision in agricuRuraI prod- 兵),SONG J(宋健).studyon starw- uct detection f机器视觉在农产品检测 berry harvesting robot:II.1mages f1 71 BURKS TF。SHEARER SA,HEATH 中的应用研究)【D】.Chongqing Univer- based identifiactions of strawberry JR.Evaluation of neural:network clas- siyt.2009. barycenter nad plucking position(草莓 sitfers for weed species discrimination. [24】WANG YL(王玉亮),LlU XX(刘贤喜), 采摘机器人的研究:¨.基于图像的草莓 【J1.Biosystems Engineering,2005,91 SU QT(苏庆堂).Maize seeds varieties 重心位置和采摘点的确定)【J】.JoumaI (3):29 304, - identification based on multi-object of China Agricultural Universiyt. 【18]WU LL(吴兰兰),LIU JY(刘剑英), feature extraction and optimized neuraI 2005,10(1):48_51. WEN YX(文友先).Weed identification network(多对象特征提取和优化神经 【31l ZHANG KL(张凯良),YANG L(杨丽), methodbasedQnSVMinthecomfield 网络的玉米种子品种识别)【J】.Transec- ZHANG TZ《张铁中).Design and ex- (基于支持伺量机的玉米田间杂草识别 tions of the Chinese Society of Agricul- periment of picking mechanism for 方法1fJ1.Transactions of the Chinese tural Engineering,2010。26(6):199-204. s ̄awberyrharvesting robot(草莓收获 Society for AgriculturaI Machinery. [251XUYM(许月明),ZHANGS(张爽). 机器人采摘执行机构设计与试验)[J1. 2009。40( :,62—166. The Appliaction of Computer Vision on Transactions of the Chinese Society for 【19】YANG.HQ(扬会清),SUN LY(孙丽燕), the Classification of Agricultural Prod- Agricultural Machinery,201 1。42(9): LlUH(刘红).Studyonthemachinevi- ucts(计算机视觉技术在农产品分级中 155-161. sion system for field weed identification ‘的应用)[J】.JoumaI of Anhui Science 【32]SUN XG(孙贤刚),wu XR(伍锡如), dhd weeding by machine(田I司杂草识 and Technology University,201 2,26 DANG×J(党选举).Design and imple- 别的机器视觉系统厦机械除草研究)【J]. (5):85-89. mentation"of appie picking robot sys・ Farm Machinery,2011(19):126-129. 【2科ZHAO W(赵巍).Appliaction of ma- tem based on visual inspection(基于视 【20】JIN XJ(金小俊).CHEN Y(陈勇),HOU chine vision on nondestructive deter- 觉检测的苹果采摘机器人系统设计与 ×G(侯学贵).Weed recognition of the urination and classifying ot wateIrmelon 实现)【J】.JoumaI fo AgdcuRural Mech- machine vision based weeding robot (机器视觉在西瓜无损检测与分级中的 anization Research。201 6(9). (基于机器视觉的除草机器人杂草识别) 应用)【D】.Huazhong Agricultural Uni- 【331 TANG YP(汤一平),WANG WY(王伟 [J1.Joumal of Shandong Universiyt of versiyt,201 3. 羊),ZHU W(朱威).Tea ridge identifi- Science and Technology fNaturaI Sci- 【27】HUZM(胡志明).Appliactionof com- action and navigation method for tea- ence Edition),2012,31(2):104 ̄1O8. puter vision classification technique in plucking machine based on machine 【21]FENG SJ(冯世杰),WANG YP(王艳 tea quality inspection(计算机视觉分级 vision(基于机器视觉的茶陇识别与采 平).Ontheinter-rowweeds recognition 技术在茶叶品质检验中的应用)[J】.Tea 茶机导航方法)[J】.Transactions of the of greenhouse crops based on ma- in Fujian,201 6。38(5):22—23. Chinese Society for AgriculturaI Ma- chine vision(基于机器视觉的温室作物 f281 ZHOU Z。BONTSTEMA J,VAN L.De- chinery。201 6。47(1):45—50. 行间杂草的识别)【J】.JoumaI of Xinyang velopme ̄t of Cucumber Harvesting 【341 LI S.Appliaction of computer vision Agricultural College,2012,22(2):115— Robot in Nether ̄nds【J1.Transactions tcehnology in tomato picking under 117. fo the,Chinese S蝴fo Agricultural naturaI backgroundfJ】.Journal of Agri— [22】WANG SW(王树文),ZHANG CL(张长 Engineering。20o. cuRuraI Mechanization Research.201 7. 利),PANG JL(房俊龙).Automatic [29】LIU CL(刘成良) WANG HY(王红永), Responsible edltor:Na LI Hepatotoxicity and mechanism of action 【8】PALMES D。SPIEGEL HU.Animal mod- of haloalkanes:Carbon tetrachloride aS els of liver regeneration[J].Biomatarials, (谭华炳).Experimental research atoxicologicaI modeI『J1.Cnt Rev Toxi— 2004。25(9):1601-1611. progress of curcumin in treating hepatic col。2003,33(2):1 05—1 36. 【9】LlU YG(刘永刚),CHEN HC(陈厚昌)。 diseases(姜黄索防治肝脏疾病实验研究 【6】WANGC(王超),Ll K(李侃),ZHANG JlANG YP(蒋毅萍).Protective effect of 进展)『J1.Chinese JoumaI of Gerontolo. YX(张燕翔),et a/.effects of quercetin curcumin on experiment liver injury in gY,201 3,33(i 1):2722—2724. solid dispersion on oxidative stress of mice(姜黄素对小鼠实验性肝损伤的保 [4】MA YG(马永贵),LUO GH(罗桂花), carbon tetrachloride inducing acute 护作用)[J】.China JoumaI of Chinese CHEN Z(陈志),et a1.Preliminary re- hepatic injury in mice(槲皮素固体分散 Materia Medica。2003。28(8):756-758. search on model stability of acute in- 体对小鼠CCI4急性肝损伤氧化应激的 【10】XU DH(许东晖),WANG S(王胜),JIN iuredIiverIeadbyCC14ofmice(小鼠四 影响1 fJ1.Modem Preventive Medicine。 J(金晶),ef aL Advances in studies on 氯化碳急性肝损伤模型稳定性初探)[J】. 201 3。40(1 5):2872—2876. pharmacological effect of curcumin(姜 JoumaI of Qinghai Normal Universiyt 【7】YE WF(叶维法).Clinical hepatology(I临 黄素的药理作用研究进展)[J】.Chinese (Natural Science Edition),2006(1):8&国1. 床肝胆病学)【M】.Tianjin:Tianjin cSience TraditoinaI and HerbaI Drugs。2005。1 1 【51 WEBER L\^,,BOLL M。SITAMPFL A. and Technology Press.1 998I (11):1737-1740.