Pattern Recognition By Interactive Optical Vision Systems in Wood Research

Pattern Recognition By Interactive Optical Vision Systems in Wood Research

( :() p ~ r ig hl © I F.\( .\Ii1(llll.t1i( \k,I'llll'rlll"rl! ,1111\ ( :0 111 1"01 ill \ \ "l' (ld\\(lrkill~ 11](111"1\ " l'II ,lIi,1.I\, 1. C/t...

1017KB Sizes 0 Downloads 17 Views

( :() p ~ r ig hl

© I F.\(

.\Ii1(llll.t1i(

\k,I'llll'rlll"rl! ,1111\

( :0 111 1"01 ill \ \ "l' (ld\\(lrkill~ 11](111"1\ "

l'II ,lIi,1.I\, 1.

C/t'd1O'\( 1\ ,d,i.t. I ~ ISti

PATTERN RECOGNITION BY INTERACTIVE OPTICAL VISION SYSTEMS IN WOOD RESEARCH

s.

Katuscak*, P. We rner*, W. Koditz** and P. Baurschmidt***

::' \ /01 ,

F UII'( /)// J(/lfrl,

1<('\ ('(1 1"{"

11I,lilllll'" I- (lIlw(\/.'(i I. Urrll i'/" "fl. L:"d'll,/u . ""kifl

"" '1\'''"/1',,11, Ii ildllllllh", (;III MI , F I/llltg ,lI fi"I/'''", FlU;

""'''''L . /,.i" Frill/n1l11ik (;II,M/, ()/I('I'/:IIII!nl. Fl U,

nbstr8ct : Tte ~ic of t~is paper was to sh ow the possibilities of using i~a ge analysis in the fie l d of measurment of text ure . Th e sample s of fi ne - li~e ve~eer and synthetic decorative foil were measured using : 1 . TV 50sch Camera: 117L i n combination with interactive system fo r image analysis I S~5 2, ~o~tron , 2 . microscope 2 eiss ~C~ provided with photometer .:;:' 8!'":C scenLi~5 table , j . trichromatic cclorimeter ::'lrepho ;,.at :FC 5 , leiss . :he ~ist ribution curves of lightness in one straight- l i ne 8~d on the sur'fec e are sho'W-n . For prac tical mea suring of col or texture it is necess8 r y to scan the texture under standar diz ed conditions acco rding to requireme n ts of the International Organization for Il l uminat i on CI~ . Ferspect ive poss ibilities of using the image analysis in wood research and industry are sho'Wn . Keywords : Image analysis, pattern recognition , optical proper t ies , lignocellulosic meterials , wood . 2 . Th e surface distribution of on e opt ical property , e . g . lightness Y or luminosity L, or so ca l led grey va l u e.

I.i\ '-:;i\ C=-jU C :~' I Oi\

Growing shortcoming of high- grade decora tive wood species f or the furn iture and wood industries stimulats the study of opt i cal properties of wood materjals and new ways and processes of ref inemen t the texture and aesthetic value of wood .

J . The surface dist ri but i on of co l or,

i . e . distr i bution of mi n i mum J valu es

e. g . hue n 0 , chrome C and li ghtness L i n en; L.:"5 - 1 976 /'lIiszecki , 1 982/ .

Opt i cal properties and their d istribution in lignocellulosic materials represen t the ccmp l ex of information . ~apid ly de veloping branch of kybernetics - tte image analysis, has enabled to d igitalize such i nfo rmation . but the use of this valuable method in the area of wood is very beginning yet .

4. The surface di str ibu t ion of co lor and / or other optical propertie s wh i ch are necessary for the comp l ex desc r ip t i on of the texture . The paper a l so int r oduces ot her po ss i bil ities of util i zing the imag e analy sis in wood research and indust r y .

The image analysis was applied in quant itative evaluation of cicr ostructure of wood species Cr yp tomeria japonica J . ~on and Chamaecyparis obtusa ~ndl . in relat i on to the f l ow rate of monodisperse d sphe r ical partic l es t hrough these wood spec i es d e scr i bed by Ohgo sh i et a 1. / 1 982a, 1982b/ . Ihe prognosis of futu r e po ssi bil i ties of image an alysis use in wood r esearch and indu s try , and a l so i t s s ignificance for quickening the advance s i n e l ect r on i cs and kybernetics app li cat i on conc ern i ng t he complex processing of wood materi al have bee~ recently discu s sed by Katu~~~k and Er~z / 1 986/ .

The fo l lowing s amp l es we re mea s ured: fineline veneer ";, r owa l nut PI" /Pr eg lejka, 2ar novica , Czechos l ovak ia/ and s yn thetic decorat i ve foi l wi th th e tex t ur e of oak wood "Jekor Liche 881" /Letron GmbH , W. Germar.y / . The fo l lowi ng s y stems of the i mag e analys i s we re u s ed: 1. Sensing of i mage by col or TV c amera 50 s ch T 1170 and s c annl ng by i n teractive s ystem for i ma ge ana l y sis l BAS 2 Kon tr on 3i l danaly s e GmbH , W. Germany/ . Lvalua t ion wa s carrie d out by using common l y applied s oftware.

Th is paper presents the possibi l i t ies of a pp l ying the image analy sis for the mea s u ri ng of wood texture on the fol l owi ng l evel s: 1. Th e l i near distribution of l i ghtn ess on the li ne perpendicular to th e f iber d ir ection .

2. Scanning by mi c r os cope ACM provided with monochr omator cont i nuou s filter 11l'1

Ill)

S. hatlls{,ik d ,,/. wit~

~

t~e

ra~ge

4GC - 70C

~m,

photo~e­

ter ~ Bn~ sCBn~ing table le. Zeiss :: eiY'. echnLc -:;~bi-i,:,. Jermany/ or, lir.e wi t!". C s::;~~.~..· er' ~~.? ~182? ... eas "J_rir.g cC!'.di tio!".s: c8J.ibr:.!tic:-. 8gei!:st m8gr.esiun: oxi~ stBr.dBr~ ar.d a£Binst the b:Bck st3r,~8 r~ ~ccnr~i~g ~o :I~ 5~~j ; §ource o! lig~t: halogen 18~p 12 V/ I OG J ; objective: Epiplen 6/C.2C ~: ; condenser: =::~ /45 ; s~~~t velocity of sC8~ning table - ...il.b IJ:ll/mir._; contint;o~s :ilter )....1 /2 = 11 Y'.m. ".eas·,.; rer::eY'.t of lightness was carriee. out with the heIr o~ the trichror::atic colori!!leter ,,"lre,,['o .,:at ~:C 5 16eiss :C. eir.tech.Y.ik Jmb~, ~. ~ercany/ . :he source of light:

: 05, /~5~0 0~/ ; diar::e ter of measuring apertu r e - 2 =; :neBs;.;.rec by Gloss trappi~e; ; sphere Geometry: (Uff;lsive il1wnination obse rvation at 8 old : 8 0/, calibrat.ior. agains t. the 5aSu star.dard /; ,.erckl 4 and against black standard accorcing to ~~~\ 5l.-jj .

Cptical ~roperties in visual rang e of the spec true are those parameters which determir.e the val ue of wooe. material in its use for decorative purposes. 7h e optical prop erties are namely as follows: - calor and its partial tristimular components; - distribution of the color parameters on visible surface and "in situ"; - gloss, du llness, opacity, transparenc y ; - stabilities, or kinetic function of visual optical parameters characterizing changes due to the effect of the light, chemical, physical and biological effects in the modification process, further in consequential technological operations Bnd in the practical use. The lir.ear distribution of lightness 7he distribution of lightness perpendicularl y to the fiber direction is the most simple procedure to express the texture. ~s an example figure 1 presents the lightness linear distribution curve of fine-line veneer "".rowalnut P 1". The lightness of specimen is expressed in relative units of grey value on the axis y. The lightness may be calibrated by means of the white and black standards. Figure 2 introduces the lightness distribution curve of the synthetic foil with the texture of oak wood. Figures 1 and 2 show that the texture of the fine -li ne veneer sample is finer than that of the oak wood like synthetic foil. Th e average variation of lightness may be expressed statistically e. g. by the distances between the minima or maxima on the axis x. It is also possible to see from observed patterns that the average lightness of fine-line veneer specimen is higher than the lightness of synthetic foil. The average colorimetrically measured value of lightness equals Y 39.42 in the case of the synthetic foil and

=

=

20.6C for t h e fi~e -li~e ve~eer i....::"3 1 '3 701 • .,.uLlal relative diffe rences of lightness betwee~ the light B~~ dark parts of the foil rerrese~t i~ the synthetic foil "I·e:-<:or i:..iche Ecl" arproximately 3D ~ , whereas in the case of fine lir..e veneer ";'.r ov;alr.'J. t r"l" approxiDately 15 ~ of the lightness scale.

Y

IC:;::::

gi ven simplest way of linear pa tter n perpendicular to f i b er ~irec ­ t.ion enables to meas~re the ~istribution of lightness in one straight-line o~ specimen surface. It is suitable mainly for the ~efinition and digitalization of the patt9 ~ "S of woc~ materials wit h relatively simple ra~i.al irai~-.• Ihe sr.cwn textt.
~he

reco gn itio~

Sur fa ce d istr ibution of li6 htness Figure J shows surface d istr ibution of lightness of t.~e syr.t.hetic foil with the texture of oak wo od . I he lightr.ess of the specimen is eypressed on ayis y similarly as in F i gs . 1 and 2 by grey value relati ve un its. 7 he axis x corresponds to the distance in the ~irec tio n perpendicular to the fiber, while the axis z is identical with the fiber d i rec tion. ihis image technique enables to express the texture of woo d surface bv set of numbers. Sta tistical evaluati~n of this set of numbers can be found in Fig . 4 . F igures 1 - 3 show the measurements of the grey values. The refore, this type of measuring expresses to a certain degree only the black- and-white ima ge of texture, that means, it corresponds to blackand -white vision t o a certain degree . Color texture The definite procedure of measuring the color texture has not been de velo p e ~ until now. ~ t the present time it is possible to measure objectively the color of homogeneous or.e - color surfaces. :he wood color phenomenon is described probably in most complex form b;y existing Czecho slovak interactive r-:OlO",=; ;" :::.~.B;'.l';~;, - S:JvJ at State Forest ? roducts ~es earch Institute 13Iv~/ , 3ratislava. This contains up to now colorimetric description of 750 wood species IKatu§~~k, 1 986 ; Eu rda, 1 986/ . Yet in this system the average color of wood is expresse d as the color of homogeneous one color ob~ect with coresponding statistical characteristics.

7 he complexity of objective color texture expression is indicate~ already by Figs. 1 - 3. Eac~ ~ int of that curves represents one point on the surface of wood /black and white vision/. In color vision eac~ color point may be objectively des cribed by the color spectrum. ~s an example the total description of two color points on fine -li ne veneer ~rowalnut P 1 Ifrom Fig. 11 is shown or. F ig 5. It represer.ts measuring places at the maximum level/curve 1/ and minimum level

Pattern Recognition by Interacti\"e Optical Vision Systems

/curve 2/ of lightness. From reflection curves on Fig. 5 it is possible to calculate characteristic tristimulus parameters for different ways of colorimetric description of calor at standardized light sources /.:..., C, D 65, E/ (\'iyszec;Ci, 1982; raulinyova, 1 981/. :he given problem may be simplified if each point of distribution curves ~ s characterized in color expression minimally b;y three tristimulus variables. It is r.ecessary for practical measuring of color texture to scan the texture under standardized conditions using three filters according to requirerrer.ts of the International Organization for Illuminat~on CIE /Commission I n ternationale de 1 Eclairage/. It is supposed that further development of analysis of color texture by applying the image analysis will proceed in the direction ir.dicated above. rtespective possibilities of using the image analysis in wood research and industry are as follows: 1. The formation of the database of textures and other optical properties of natural wood, composite wood materials, impregnated papers and foils and other materials of wood processing industry. 2. Objective identification of wood species.

J. Scanning microphotometric analysis /Si';'A/ .

4. Objective evaluation and interpreta-

tion of results y ielded by the electror. and optic microscopies and macrophotographs.

5. Objective measuring and evaluation of biological state of tree substance. 6. i.:odelling the complexes of optical properties determining the aesthetic sensation.

7.

checking and evaluation of the appearance of wood-processing industry products.

~uality

8. Time-sequence relationships of the optical properties evolution of products manufactured by wood-processing industry and the prediction of future developmen t. 9. The study and development of new processes of chemical, physical, biochemical and biological texturization or modification of wood. 10. Automation and control of technological processes of wood-processing industry utilizing the pattern recognition. CONCLUSION There are three principal possibilities of applying the image analysis to measure the texture of wood: /1/ the linear distribution of lightness, /2/ surface distribution of one optical property and /3/ surface distribution of color as a mini-

III

mum tristimular parameter. These texture features and differences between various textures can be digitalized, quantified, stored and compared. This can be valuable tool for research and development of texturization and other wood modification processes, for development of new aesthetic wood products and for control of woodworking processes. 2EF El-,El\C 1S Hurda, 3. /198 6/. Hognoceni prirozeneho odstinu barvy dreva, Drevo. in press. hatuscak, S., Durisova, L. / 1986/. Opticke vlastnosti dreva vo viditernej oblasti spektra, Kolordatabanka ;jDvU. Drevo, in press. i<.atusc ak, S., Hrllz, 3 . / 198 6/. l"oznosti vyuzitia analjzy obrazu v dreverskom vyskume a priemysle. Drevo. 41 /1986/ 131.

Ohgoshi, ;,i ., hakato, i., Sadoh, T. /1982/. ~uantitative Evaluation of BorderedPi t ["i.embrane Pores of Softwoods by the Image-Analysis System. Je Jape Wood Res. Soc., 28 /1982/ 583. Ohgoshi, ~ ., Nakato, K., Sadoh, T. /1982/. Contribution of Bordered-Pit Membrane Pores to Flow through Softwoods. J. Jap. Wood Ftes. Soc. 28, /1982/

.2.2£h Paulinyova, E., Katuscak, S. /1981/. Frogramy na objektivne hodnotenie farebnosti, farebnych diferencii a egality farby z reflexnych spektier. Drevtrsky vjskum 26 /1981/ JJ. Wyszecki, G., Stiles, W. S. /1982/. Color Sc i enc e-C onc epts and illethods, ~uan­ titative Data and Formulae, 2 - nd Edition, John Wiley and Sons, New York 1982.

11 2

S. Ka tll sc-;\ k

1'1 Ill.

GF·EYI.}A LIJE ·.ce

£.. .) .)

1em

PO I tiTS P1,. 1. LipU. •• (1.tr1~ut1 .. eurt. of fille-la. T•••• r "4zeIral.Ilut P 1" In-.cl.Jka, C•• ..-.doTakia/ S•••1.IIg of ~ge b1 oolor rv _era Bo.oII 'r 1170 ul .oazuo1ag b7 at.reo 1 ... .,..h. for u.ge .....17d. IB4S 2 IKo.troa, 11. Ger-

t

~/.

I;

F'EY1.1 ALIJ E

400 PO I tiT ::; Pl,. 2. L1pta••• '1.tr1but1. . curT. of ~th.t10 d.oorat1Te f011 with the texture of oak •• 04 ILam., 11. G.~/. S•• e1a& of t.age b1 oolor 'rV oeaera Boach or '170 . . . .~ ~1 ilIt.raotiT• .,.et. . for t.age aaal;re1. IB4S 2 / Kontro n, 11, G.~/.

Pie. ). Lipu. .. di.tribuUon CurT" of the e:rnth,Uo fo11 ..... f.c. WiU. the texture of oak wo04 - graph.1o ropro ... taUon of ~laot-"""'wlUt. Th1 .. of oak texture. S.ae1As of 1Mgo b1 oolor orv _ r a Bo . .1I or 1170 . ., 111 ut.rectin .,..t. for 1Mgo anal;rd. IBlS 2 lIo.b'oD, 11. Ge:rwaa;r/.

.-as

11 3

Patlern Re co){nitio n bl' Int e rartin' Optical \ ' ision S\'st e ms

,-.

,-.

,-.

.:- TATI .:. TI C.:,

Gf'EY"L

11EH11 "HLIJE 1~ I :: ~~ I ~ I) I :' 1';.(' ,:.::: I ~ 2 Ot. I ~ 2 :: ~~ 1';,(1 «. ' ) 1t.1 (1:1';·1 ~) g 1';·2 45 1':,2 l)O

I', I~'

' "i

.....

. .. . ·· ... · .~--. ... .--.-.-,.....---.....----~.

~

.'-. ..

-.. -+

i

I

1[1

XR

-:.I [I

[lE" 14 14 14 13 12 12 12 14 14 13

MH :<

MIH

..-,

~

61

t·:· 2t: ~I 6~

7::! 4::: {:::

35

12t. 12: 126 125 125 11 ::: I'"co I;", 121 124

.

19::: 1"3 1"4 I :~.; 193 195 I" ;'

~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~ ~~

2~2 2~1

191

; I

1

2~

No,

Color1_ trio ftlue.

I

::Iouroe • S. .pIe 1 .. ------42,9) 12,05 X y

z

80

. x

1

6()

I

D 65 1 )4,16

9,2) 8,67 5.16

)6.19 »,24 20,31

4,12

• J9ElO ,)816 .)610 ,)684 .2)50 .2500

.4(0)

.40)2 .)704 ,2264

,422) ,Jn1 .2006

)5.50 6),86 1.58 9,99

)5,'. 8.5) 16.16

64.)6 9.94

)5.75 8.57 16,9)

9,4~

)5,44 1,20

9,6) , .68

32,88 24.01

8,15 5.60

,50 1 7

,5159

,)19U ,)585 ,2624 6.,U6 8,76

,.'42

,4120

,08.41

,0121

66,09

J1,16

a

11,11

10,2)

b

;:4.12

1 9 .30

20.)8

»,19 )2,6J 22,14

16.41 20,09

,)160

.22)7

20,98

9,94

8,88

I

40 ~

,'2

./

/

20

-- - _._---_.- .. 450

soo

550

600

./

,/ /

650

~ IIm

700