小麦生长信息计算机视觉检测技术研究_英文_

小麦生长信息计算机视觉检测技术研究_英文_ 农 业 工 程 学 报J an. 17 N o. 1 第 17 卷 第 1 期 136 T ran sac t io n s o f th e C SA E J an. 2001 2001 年1 月 () : 100226819 20010120136204A r t ic le ID Com pu ter V is ion D e tec t in g Sy stem f or W hea t Grow th In f orm a t ion , , , , , G e ng N a n He D o ng jia n W a ng J ing D a ng G e ro ng W a ng Ho ng b in G uo M e ng s he ng (, C ol leg e of M ech an ica l & E lec t ron ic E ng inee r ing N or thw es t S c ience T ech nology ), 712100, U n iv e rs ity of A g r icu l tu re and F ores t ry Y ang l ing C h ina A bstrac t: T h is p ap e r fo cu sed o n th e tech n ique o f w h ea t g row ing info rm a t io n m ea su rem en t by com p u te r v isio n. , , B a sed o n w h ea t leaf g row ing p ecu lia r it ie su sing th e num be r o f fo rk s be tw een leave s and stem s af te r th inn ingth e . num be r o f leave s can be o b ta inedT h e re sea rch def ined th e w h ea t leaf co lo r w ith h ue and m ea su red th e co lo r o f . , + + .leaf by h ueA s a re su ltth e so f tw a re w a s deve lop ed w ith V isua l C o n W indow sy stem : ; ; ; + +Key wordsw h ea tcom p u te r v isio nim age p ro ce ssingV C CL C n um ber: T P 391. 41 D ocum en t code: A accu racy o f exp e r im en t s can no t b e en su red. 1 In troduc t ion 2T h e m ea su rem en t o n w h ea t g row th in fo rm a t io n th ro u gh com p u te r v isio n is an u n to u ch ed T h e c lo sed co n t ro l o ve r p lan t g row th sy stem , . , , de tec t io n w ith h igh accu racyIn add it io n w e can can m a in ta in a ll liv in g co n d it io n s u n de r b e st sta te 2e stab lish a w h ea tg row in g m o de l an d m ak e an , im p ro ve p ro du c t iv ity an d y ie ld t rem en do u sly , an a ly sis d irec t ly o n th e da ta co llec tedo ffe r a w h ich is an in ev itab le t ren d o f in te rn a t io n a l 1, 3, 4 . . , re liab le tech n ica l suppo r t fo r w h ea t b reed in g ag r icu ltu re deve lopm en tT o rea lize itin fo rm a t io n . abo u t p lan t s n eed s to b e co llec tedT h e p ap e r 2 D e tec t in g Sy stem in teg ra te s th e tech no lo gy o f com p u te r v isio n w ith . gen e t ic en g in ee r in g to ge th e rT h e de tec ted o b jec t A s F ig. 1 show s, th e sy stem co n sist s o f a PC ( ) 300, IBM A co n n ec ted w ith D ah en g Im age is w h ea t. B a sed o n an an a ly sis o f th e o b ta in ed , Com p an y CCD p u re co lo r g ru b b e r bo a rdR eco rd w h ea t re sea rch acqu ire s th e im age s, th e tu rn in g C am e ra, illum in a t in g dev ice, tab le. in fo rm a t io n abo u t co lo r an d n um b e r o f leave s. C P 410 Co lo r CCD C am e ra, m a tch ed P an a so n ic Som e exp e r im en t s a re u n de r tak en to illu st ra te it s 12122. w ith D H C L a se rT h is is u sed to o b ta in tw o.fea sib ility 32. qu ad ra tu re p ic tu re s fo r D g row th in fo rm a t io nA S in ce w h ea t ow n s life, in fo rm a t io n abo u t th e , . tu rn in g tab le ca r r ie s sam p le san d ro lls a ro u n dA n um b e r an d co lo r o f leave s can b e o n ly o b ta in ed cam e ra co u ld tak e tw o p ic tu re s in tw o d iffe ren t , , , m an u a lly a t in te rva lsw h ich f ir st lyrequ ire s a . d irec t io n sT o m ak e th e ligh t ca st even ly o n to th e , , la rge am o u n t o f m o n eym a te r ia ls an d labo ran d , 602o b jec t s w itho u t an y ref lec t io n six W lam p s a re ; , th e eff ic ien cy is low seco n d lyit is su b jec t to , som e su b jec t ive fac to r sth e re su lt s a re d iff icu lt to b e co n sisten t w ith each o th e r; th ird ly, th e m ea su rem en t is a to u ch ed o n e, w h ea t m ay rece ive , d irec t p re ss an d ch an ge po stu re a s w e llth e 210225: 2000R ece ived da te 1. PC 2. G rubbe r bo a rd 3. CCD C am e ra 4. T u rn ing tab le : Fo unda t io n item N a tu ra l Sc ience Fo unda t io n o f Sh aanx i P ro v ince 5. 6. 7. W h ea t sam p le Illum inan t B ack g ro und ( )4130103 : , , ; , , B io g rap h ie sGeng N an m a lelec tu re rH e D o ng jian m a le. 1 F igCom p u te r v isio n sy stem . . , .P hDp ro fe sso r ?60 f ix ed eve ry a ro u n d a t th e bo t tom o f th e p la te to .illum in a te 30 w h ea t s w e re m ea su red in th e exp e r im en t , af te r e lo n ga t io n stageth e se sam p le s w e re p lan ted o n O c to b e r 1999. Fo r th e co n ven ien ce o f m ea su rem en t, seve ra l de lica te w h ea t s w e re tak en w ith o n ly o n e w h ea t lef t in o n e cu lt iva te aw ay, . 2000, p la teIn A p r il th e m ea su rem en t w a s F ig. 4 W h ea t th inn ing im age .sta r ted 30 w h ea t s w e re p u t o n th e p la te 8 T h e n e igh bo r in g va lu e s o f eve ry p ix e l w e re 2 re sp ec t ive ly an d m ea su red th ro u gh th e se lf , 0 7 , an d m ak e D a ta sto red in a r ray D a ta . deve lop ed so f tw a re sy stemIn fo rm a t io n abo u t th e 2C ou n t T h en th e e igh tva lu e D a ta 8 = D a ta 0 . , .n um b e r o f leave sco lo r s an d len g th w e re acqu ired can b e ca lcu la ted a s: 7 3 Pr in c ip le s f or the D e tec t ion 1 ()D a ta [ i + 1 | 1 C ou n t = D a ta [ i -| ?2 i= 0 3. 1 D e tec t ion of Num ber of L ea ve s It w a s a ssum ed th a t th e o b jec t p ix e l is 1 an d . 2 .A s F igis an im age o f w h ea t af te r jo in t in g 0. , th e b ack g ro u n d p ix e l is T h e refo reit m u st b e a w h ea t af te r e lo n ga t io n is n ea r ly ve r t ica l to th e C ou n t 3. fo rk w h en th e is equ a l to A cco rd in g to g ro u n d, th e re is a ce r ta in an g le b e tw een leave s an d ( )1 fo rm u la p ro ce ss th e w ho le th in n in g im age . , stem sCo n side r in g th a t leave s g row w ho r ledeve ry fo rk can b e de tec ted. T h e n um b e r o f leave s th e re is a sin g le fo rk b e tw een eve ry leaf an d .can b e o b ta in ed af te r co u n t in g a ll fo rk s 5 . stem W e a re ab le to th in th e im age w ith f low T ab. 1 list s th e n um b e r o f leave s af te r . 3, ch a r t show n in F igan d th en th e in fo rm a t io n 30 w h ea t s in th e exp e r im en t. It a ll de tec t in g ()abo u t th e w h ea t st ru c tu re is acqu ired F ig. 4. in d ica te s th a t th ro u gh th is w ay w e can de tec t th e , 2 30n um b e r o f leave s m o re accu ra te lyo n ly o u t o f w e ren ’ t de tec ted co r rec t ly, th e p rec isen e ss ra t io is 93. 3 %. Tab. 1 The re sults of leave s n um ber de tec t ion M ea su red R ea lM ea su red R ea l N o.N o. va lue va lueva lue va lue 1 3 3 16 3 3 2 4 4 17 3 3 3 4 4 18 2 2 4 4 4 19 5 3 5 4 4 20 2 2 6 4 4 21 2 2 F ig. 2 W h ea t im age af te r jo in t ing 7 2 2 22 2 2 8 4 4 23 2 2 9 2 2 24 3 3 10 3 3 25 5 3 11 2 2 26 2 2 12 3 3 27 2 2 13 2 2 28 3 3 14 3 2 29 2 2 15 2 230 2 2 3. 2 Co lor D e tec t ion F ig. 3 F low ch a r t o f de tec t ing th e num be r o f leave s T h e co lo r o f a leaf is a com b in a t io n o f va r io u s , co lo r sth e ra t io b e tw een ch lo rop h y ll fo r g reen an d 农业工程学报 138 2001 年 ca ro teno id fo r ye llow de te rm in e s th e o ve ra ll co lo r. H e re, w h a t w e a re co n ce rn ed abo u t is va lu e s , . V a r io u s c ircum stan ce s affec t th e co lo r o f leave so f h u e sT ak in g th e to ta l n um b e r o f p ix e ls o f a ll , , , su ch a s su n sh in etem p e ra tu rem in e ra ls an d w a te r leave s o n a w h ea t stem a s sam p le sw e ge t th e ; , . in so ilth e refo reth e co lo r o f leave s is a ve ry ave rage o f a ll va lu e s an d stan da rd to le ran ceIn ., .sen sit ive in dex fo r n u t r it io n an d h ea lth o f p lan t s th is w ayco lo r ch a rac te r ist ic s can b e de sc r ib ed . 6 . 28 Seve ra l co lo r sy stem s co u ld b e u sed to F igis p ix e ls d ist r ib u t io n o f w h ea t leaf N oin . , . , de sc r ib e co lo r s in n a tu reIn gen e ra lth e re a re th e exp e r im en tA cco rd in g to th e re su lt sn ea r ly (() , , , H L S , , 225?,R GB R ed G reen B lu eH u eL igh tn e ssa ll h u e s d ist r ib u te a ro u n d th is is o f g rea t ()CM YB , , , an d C yan M agen taY e llow h e lp fo r fu r th e r re sea rch o n th e re la t io n b e tw een Sa tu ra t io n ). R GB , .B lack p re sen t s co lo r th ro u gh m ix in g red h u e s an d w h ea t liv in g co n d it io n s , , g reen b lu e to ge th e r in d iffe ren t ra t io so f ten u sed H L S in d isp lay in g sy stem. can de sc r ib e co lo r s in te rm s o f qu an t ity. It u se s h u e s to show co lo r in ; o rde r to d ist in gu ish th is co lo r f rom ano th e rit u se s , sa tu ra t io n fo r co lo r th ick n e ssligh tn e ss fo r th e .b r igh t sen se o f co lo r S in ce th e re sea rch n eed s to de sc r ib e leaf co lo r H L S in te rm s o f qu an t ity, is app lied to de sc r ib e co lo r. T h e g ru b b e r bo a rd co llec ted p u re co lo r . R GBim age s b efo re de tec t io nW e sho u ld co n ve r t H L S . to T h e re la t io n b e tw een th e tw o sy stem s is F ig. 6 H isto g ram o f p ixe l d ist r ibu t io n o f w h ea t leaf :illu st ra ted a s fo llow s - 1 4 D e tec t in g Sof tware D e s ign ) ((H = tan [ 0. 7R - 0. 59G - B ƒ- 0. 110. 3R - ) () 0. 59G + 0. 89B ] 22, B a sed o n th e abo vem en t io n ed th eo r ie sa ()3 L = 0. 3R + 0. 59G + 0. 11B com p u te r v isio n de tec t in g sy stem fo r w h ea t g row th S = in fo rm a t io n w a s deve lop ed w ith M ic ro so f t V isu a l C 2 2 2+ + 98.() ) o n Pw in (0. R 7- 0. 5G 9- 0. B11 + - 0R . -3 0. 5G9 + 0. 8B9 ()T h e sy stem h a s a goo d m en u fo r th e w ho le 4 . p ro ce ssT h e u se r’ s in te rface loo k s a t t rac t ive an d W h en de tec t in g co lo r, f ir st, w e sep a ra te . can b e co n t ro lled ea sily an d co n ven ien t lyF ile s an d leave s f rom stem s b a sed o n th e po sit io n o f fo rk s in ; da ta m an agem en t is qu ite effec t iveth e o r ig in a l , an im ageth en co n ve r t eve ry p ix e l o f leave s in to a da ta a re ea sily p rep a red. T h e sy stem h a s a ( ) ( ) ( )H L S 2, 3, 4. va lu e in u sin g equ a t io n T h e pow e rfu l fu n c t io n o f a lte rn a t io n , an d is ab le to p ro ce ss f low ch a r t is show n in F ig. 5. . g ive a id s an d p e rm it s to le ran ce in th e p ro ce ssT h e ; sy stem is u n ive r sa l to som e ex ten tit is th e sam e u sefu l fo r o th e r c rop s com p u te r de tec t io n su ch a s co rn an d r ice. T h e sy stem h a s an ex ten ded in te rface in w h ich C Im age Po in t P ro ce ss an d , C Im age A rea P ro ce ss c la ss w e re exp lo redan d it co u ld accom p lish b in a ry, sm oo th in g, th in n in g an d 2.T h e sy stem co u ld b e o th e r im agep ro ce sse s rev ised th ro u gh c la ss in h e r itan ce an d c la ss F ig. 5 F low ch a r t o f co lo r de tec t io n de r iva t io n to f in ish th e exp an sio n to th e sy stem. 5 6 Pro spec ts Con c lus ion s T h e re sea rch is to e stab lish com p u te r v isio n T h e re sea rch in teg ra te s va r io u s su b jec t s de tec t in g sy stem fo r w h ea t g row th in fo rm a t io n in ., to ge th e r Fo r th e f ir st t im eit adop t s com p u te r . b reed in g p ro ce ssT h e h a rdw a re can b e d ispo sed in v isio n fo r dyn am ic rea l t im e no n co n tac t de tec t io n in .tw o w ay s . th e p ro ce ss o f w h ea t b reed in gIt is o f sign if ican ce ) . 1. 1, in th e th eo ryT h e f ru it s o f th e re sea rch a re g rea t ly J u st a s show ed in F igth ro u gh h a rdw a re , p rac t icab le fo r p rop e r an d ra t io n a l m an agem en t o f an d so f tw a re sy stem w e can co n t ro l th e tu rn in g o f , . tab le w ith sam p le st ran sfe r th e cu lt iva t in g p la te b reed in g an d cu lt iva t io nA n d th ey can p rom o te . to th e sam p le tab le fo r im age co llec t io nT h is th e t ran sfe r o f t rad it io n a l fa rm in g to sc ien t if ic , , m ean s le ss in ve stm en tb u t isn ’ t fea sib le eno u gh .fa rm in g .in co n ven ien t to b e op e ra ted ) 2 2 T h e au to cam e ra co llec t s im age s th ro u ghRef eren ce s , 1 , . so f tw a rean d w e a re ab le to co n t ro l th e cam e ra T a rbe ll K A R e id J FA com p u te r v isio n sy stem fo r .ch a rac te r izing co rn g row th and deve lopm en t , , .ho r izo n ta llyve r t ica llya s w e ll a s fo ca l d istan ce () T ran sac t io n s o f th e A SA E , 1991, 34 5: 2245, 2255. It is fea sib le an d co n ven ien t b u t requ ire s m o re 2 , .+ + . 2D av id J K rug lin sk i P ro g ramm ing V isua l C M i . , in ve stm en tF u r th e rm o reit co u ld accom p lish f ie ld R ?. 1998. c ro so f t P re ssim age o b ta inm en t b a sed o n In te rn e t W eb af te r th e ,, . 23 Zh ang J ianp ingW u Sho uy iF ang M ing ruQ uan t ita t i fu lf illm en t o f com p u te r v isio n de tec t in g h a rdw a re ve in sp ec t io n o f ex te rna l qua lity ch a rac te r s o f to bacoo .fo r p lan t s g row th in fo rm a t io n ( ) . , 1996, 12 3 : 158,leafT ran sac t io n s o f th e C SA E o f so f tw a re an d A f te r th e e stab lishm en t 162. h a rdw a re sy stem fo r p lan t g row th in fo rm a t io n , , , . 24 H e D o ng jian Yang Q ingX ue Sh aop inge t a lCom p u 2, te r v isio n m e tho d fo r m ea su r ing defec t a rea o f sp h e re m ea su rem en t th ro u gh com p u te rth e im age s w ith ( ) ,. , 1997, 13 4: 156f ru it sT ran sac t io n s o f th e C SA E 3rea l t im e a t D in fo rm a t io n co u ld b e co llec ted a t 160. in te rva ls. T h e de sign ed so f tw a re can m ak e an . . : 25 Yu So ng lieC rop p lan t ingB e ijingC h ina A g r icu l , an a ly sis o n th e in fo rm a t io n e stab lish g row in g . 1992.tu ra l P re ss , m o de ls an d da tab a seo ffe r th e re liab le an d .accu ra te tech n ica l suppo r t fo r w h ea t b reed in g 小麦生长信息计算机视觉检测技术研究 耿楠 何东健王婧党革荣王宏斌 郭孟生 () 西北农林科技大学机械与电子工程学院, 杨凌 712100 摘 要: 以计算机视觉检测小麦生长信息为目的, 建立了计算机视觉系统以获取小麦图像。在分析小麦生长特性的基础上, 提出了用小麦叶片和茎秆之间的叉点数来检测叶数; 用色相示小麦叶片的颜色, 对小麦叶片的色相值进 行了检测。并在中文平台上用开发研制了检测软件。98 + + 6. 0 W indow s V isua l C 关键词: 小麦; 计算机视觉; 图像处理; + +V C