Document downloded from: http://hdl.hndle.net/10251/38948 This pper must be cited s: Ivorr Mrtínez, E.; Verdú Amt, S.; Sánchez Slmerón, AJ.; Brt Bvier, JM.; Gru Meló, R. (2014). Continuous monitoring of bred dough fermenttion using 3D vision Structured Light technique. Journl of Food Engineering. 130:8-13. doi:10.1016/j.jfoodeng.2013.12.031. The finl publiction is vilble t http://dx.doi.org/10.1016/j.jfoodeng.2013.12.031 Copyright Elsevier
1 2 3 4 Continuous monitoring of bred dough fermenttion using 3D vision Structured Light technique Eugenio Ivorr 2, Smuel Verdú 1*, Antonio J. Sánchez 2, Jose M. Brt 1, Rúl Gru 1 5 6 7 1 Deprtmento de Tecnologí de Alimentos. Universidd Politècnic de Vlènci, Spin. 8 9 2 Deprtmento de Ingenierí de Sistems y Automátic, Universidd Politècnic de Vlènci, Spin 10 11 12 13 14 15 16 17 18 19 20 21 22 Abstrct The im of this work ws to monitor the fermenttion phse of bred-mking using 3D vision system bsed on Structured Light (SL). The evolution of the dough ws studied employing 10 whet flours with non-physicochemicl nd rheologicl differences. However, differences in dough behviours during fermenttion were found bsed on SL method prmeters. When the vrition of the totl trnsversl re ws relted to the mximum height t ech fermenttion time set of peks nd vlleys ppered. These sets re directly relted to the fermenttion cpcity. Specificlly, lower number of peks, during the min fermenttion time (100 minutes), re relted to whet flours with high fermenttion cpcity. Consequently, the proposed SL Technique could be used s method to check the fermenttion cpcity of whet flours ccording to their fermenttion behviour. 23 24 Keywords: Structured light, monitoring, fermenttion, bred, behviour.
1. Introduction Severl importnt fctors ffect productivity in the bred industry due to the modifictions of whet flour properties nd behviour during the bred-mking process. Some of them re: the cultivtion method, whet vriety, phytohelth products, environmentl fctors, climtic conditions, plgues, store kernel ltertions nd milling, etc., ll of which result in chnges in flour composition (Cocchi et l., 2005). Therefore, it is importnt to develop methods to study whet flour properties nd process phses to decide the best use for ech btch of rw mterils, nd in turn to modify the process prmeters when necessry. Fermenttion of the dough is n importnt phse in the bred-mking process which ffects prmeters in the finl product, such s texture, pltbility nd generl qulity. This is n importnt temperture-dependent phse, in which the metbolism of yests trnsforms ssimilble crbohydrtes nd mino cids into crbon dioxide nd ethyl lcohol s the principl end products (Birch et l., 2013). Gluten plys crucil role in creting the dough structure nd bking the bred. It ffects the stbility of the dough nd bred volume by forming the skeleton of whet dough which combines the remining ingredients nd dditives within the dough (Brk et l., 2013). The oxidtion of cysteine mino cids from gluten proteins (glidines nd glutenines) by tiol groups genertes viscoelstic network which is cpble of retining crbon dioxide from which the gs cells develop. The growth of gs cells depends on the cell size nd the dough composition (flour, wter nd other ingredients). Severl compounds re known to exert stbilizing influence nd retrd unwnted phenomen such s colescence (Gn et l., 1995). As result, dough composition nd yest ctivity re mnifested in dough bubble sizes nd dough volume expnsion. Recently, empiricl nd physicochemicl techniques hve been used to chrcterize the different phses of the bred-mking process (Dobrszczyk & Morgenstern, 2003), the mjority of them bsed on destructive nlysis. In prticulr, the fermenttion phse hs been extensively studied from different points of view by vrious non-destructive techniques (Lssoued et l., 2007). All of them re imed t obtining informtion bout the implicted fermenttion nd bking prmeters, thereby explining the process phenomen nd improving knowledge s well s control over the finl product. The evolution of prmeters such s dough volume, density nd gs cell sizes re importnt control vribles, since their behviour will hve n importnt influence on the qulity of the finl product. Imge nlysis is n importnt tool in the chrcteriztion of the bred-mking process, which hs been demonstrted to be n importnt reserch nd industril ppliction (Clderón-Domínguez et l., 2008). Different techniques nd methods bsed on multiple principles hve been used to cquire nd nlyse imges obtined during the process. Some exmples re: Confocl lser scnning microscopy (Jekle et l., 2011, Updhyy et l. 2012), mgnetic resonnce (Frnci & Serš., 2011) nd methods bsed on 2D (Pour-Dmnb et l., 2011). The structured-light method is nother imging technique. It is bsed on the projection of pttern of light on smple nd the clcultion of 3D dimensions from the deformtion of the pttern using cmer (Verdú et l., 2013). This technique permits the monitoring of continuous processes nd could be pplied on-line. Accordingly, the im of this work ws to monitor bred dough fermenttion of different whet flour smples with developed computer vision bsed on Structured Light, in order to obtin useful informtion bout the process nd chrcterize the response of the rw mteril. 2. Mteril & methods 2.1. Dough preprtion nd fermenttion process Ten different doughs were mde employing the following percentges: 56% whet flour, 35% wter, 2% sunflower oil, 2% commercil pressed yest (Scchromyces cerevisie), 4% white sugr nd 1.5% NCl. All doughs were mde using the sme procedure but employing ten different btches of whet flours which were obtined from locl fctory. Whet flour
moisture, gluten, flling number nd rheologicl prmeters were determined ccording to the methods of the Interntionl Assocition for Cerel Science nd Technology (ICC) nd protein content by the Kjeldhl Method.Tble 1 lists the verge nd stndrd devition of the prmeters vluted. Tble 1. Averge nd stndrd devition of protein, lveogrph prmeters (P=mximum pressure (mm), L=extensibility (mm); W=strength (J -4 ), flling number, moisture, gluten nd drygluten for the 10 btches of whet flour employed. P 97 ± 2 L 104 ± 3 W 373 ± 12 P/L 0,93 ± 0,03 Flling number 413 ± 6 Moisture (%) 14,28 ± 0,24 Wet Gluten (%) 31,23 ± 0,25 Dry gluten (%) 13,51 ± 0,11 The doughs were mde by combining ll ingredients in food mixer (Thermomix TM31, Vorwerk, Germny) ccording to the following procedure. In the first step, liquid components (wter nd oil), sugr nd NCl were mixed for 4 minutes t 37 C. The pressed yest ws dded in the next step nd mixed t the sme temperture for 30 seconds. Finlly, the flour ws dded nd mixed with the rest of the ingredients using specil bred dough mixing progrm which provides homogeneous dough. Then, 450 g of dough ws plced in the metl mould (8x8x30cm) for fermenttion. This process ws crried out in chmber with controlled humidity nd temperture (KBF720, Binder, Tuttlingen, Germny), where 3D Structured Light (SL) device ws developed nd clibrted. The conditions of the fermenttion process were 37 C nd 90 % of RH. The smples were fermented until the dough lost its stbility nd size, when growth depletion occurred. For ech dough, four replictes were used. 2.2. Fermenttion monitoring by Structured Light method (SL) 2.3. A 3D vision system ws designed specificlly to monitor fermenttion. This vision system ws composed of structured light nd cmer. The structured light ws generted employing red linel lser (Lsiris SNF 410, Coherent Inc. Snt Clr, Cliforni (USA)) nd the network grycmer, with index protection of 67 (IP67), for the imge cquisition (In-Sight 5100, Cognex, Boston, Msschusetts (USA)), both of them were instlled inside the fermenttion chmber (Fig, 1). The lser hd n ngle β of 0.65 rdins (Fig 1) which in combintion with the resolution of the cmer (640x480) nd its distnce from the smple give Z resolution of 1.4*10-4 m nd X resolution of 2.1*10-4 m. This configurtion ws estblished to chieve working rnge of 0.1 m in the X xis nd of 0.08m in the Z xis. The cmer worked with n cquisition rte of 1 fps due to the long period of time for the test performed (round 2 hours) but it cn work t up to 60fps. Clibrtion of the equipment ws relized by tking 10 regulrly distributed points in 3D in the lser projection plne (Trobin et l., 1995). Using these points with known 3D coordintes nd their correspondent points in the imge, homogrphy trnsformtion ws clculted (Zhng et l., 2000). Lser Lser Smple Cmer Smple Y 0 Y 1 Figure 1. Developed 3D vision system instlled in the fermenttion chmber nd schemtic 3D vision system. 2.4. SL method informtion extrction The lser points projected on the imge were extrcted following these steps: first, the imge ws segmented using Otsu s b Z
globl threshold (Otsu et l., 1979), then the imge ws filtered removing non connected pixels with n re lower thn 100px nd finlly, row coordintes were clculted by weight men. This weight men vlue ws clculted for ech column using the intensity vlue in order to get subpixel precision. The 3D coordintes were then clculted using the homogrphy from these pixel coordintes. The lst step ws to pply rottion mtrix to them in order to mke the z xis norml to the surfce s cn be seen in the reference coordinte system of Fig. 1. The smple ws defined s 3D curve composed of the 3D points which re between the known 3D points of the mould s borders. The following informtion bout the growth of the smples during the fermenttion ws cquired from ech imge: Mximum height (H): The mximum Z vlue for the smple nd its position. Trnsversl re (A): The integrtion of the Z vlues long the X direction of the smple. Arc correltion: A Person s correltion between theoreticl rc defined by two points (the two extreme sides of the smple long the X direction) nd the rdius (hlf the modulus of the vector between the two points of the rc) nd the 3D curve of the smple. Acquisition nd dt processing ws crried out using code developed in the Mtlb computtionl environment (The Mthworks, Ntick, Msschussets, USA). 3. Results & discussion Dough volume expnsion, in which CO 2 production is ssocited with yest ctivity (Bjd & Serš, 2011), ws chrcterized from dt obtined employing the structured line method. Dt were expressed s trnsversl re (A), the mximum height (H) t ech time nd the reltionship between both (A/H). Figure 2, s n exmple, shows the evolution of the dough trnsversl re with the fermenttion time. The X xis expresses the thickness of the bred, the Y xis the fermenttion time nd the Z xis the height. Figure 2. 3D representtion of dough trnsversl re evolution with fermenttion time. Figure 3 shows the evolution of the trnsversl re, for the 10 doughs tested, with ech fermenttion time (until growth depletion occurred). Although the physicochemicl nlysis of the whet flours used for mking the doughs did not show ny significnt difference (Tble 1), the times to rech the mximum trnsversl re were different for the doughs tested (Fig. 3). The mximum re vlue ws 1758.7 ± 189.2 mm 2, employing 187 minutes (dough number 10) nd the lowest ws 1027.2±53.7 mm 2 employing 108 minutes (dough number 1) (Tble 2). Similr behviour ws obtined when the mximum height t ech fermenttion time ws compred (Tble 2). When the dt bout re nd height t 100 min were compred, sttisticl differences were observed (Tble 2). The time of 100 min ws the shortest time for dough to rech its finl fermenttion nd ws similr to tht employed by other uthors s fermenttion time (Keskin et l., 2004; Bjd & Serš, 2011; Rizzello et l., 2012). Differences between doughs could be relted to the level of strch dmge, the influence of gluten proteins nd the number of bubbles nd their size (Fridi, 1990; Brrer et l.. 2013; Mills et l., 2003; Boyci et l. 2004). The frction of strch dmge is consequence of the kind nd time of milling the whet-kernels. It hs n influence on the rheologicl nd functionl properties of dough (Fridi, 1990) since it is ble to modify flow regimes, processing vribles nd swelling proprieties (Brrer et l., 2007). Excessive strch dmge could overly hydrte the dough nd permit
ccelerted enzymtic ction (Boyciet l. 2004), thus it produces differences in the vilbility of yest substrtes nd dough stbility with time. On the other hnd, climtic fluctutions (Dniel &Triboi, 2000) nd levels of severl fertilizer compounds such s nitrogen, sulphur nd phosphorus (Altenbch et l., 2002) my ffect the enzyme ctivity kinetics during whet growth, thus disturbing n optiml rtio genertion of subfrctions of gluten proteins (glidin nd glutenin). Higher glutenin levels mke the dough more elstic, thus giving the dough its property of resistnce to extension, while higher glidin content increses the extensibility of the dough (Brk et l., 2013). This rtio hs strong influence on the qulity nd technologicl proprieties of whet flour (Rdovnovic et l., 2002). However, the nlysis of strch dmge or gluten protein composition requires specific techniques nd devices tht re not esy to crry out on dily bsis in the industry. re (mm 2 ) 2000 1800 1600 1400 1200 1000 800 600 400 200 Doughs 1 2 3 4 5 6 7 8 9 10 0 0 20 40 60 80 100 120 140 160 180 200 t (min) Figure 3. Evolution of the dough s trnsversl re with fermenttion time (until growth depletion occurred). Tble 2. Trnsversl re nd the mximum height of the tested doughs, t the end of ech fermenttion time (until no increse is observed) nd t 100 minutes (time required for the dough which reched its finl fermenttion first). Dough Finl time (Ft) (min) 1 108 ± 2 2 106 ± 4 3 100 ± 4 4 138 ± 3 5 142 ± 5 6 123 ± 17 7 133 ± 4 8 155 ± 1 9 159 ± 3 10 187 ± 8 Until the end of fermenttion time Trnsversl re (A) (mm 2 ) c cd b de e f 1027.2 ± 53.7 1055.6 ± 75.5 1228.7 ± 32.3 1402.7 ± 68.4 1420.5 ± 147.8 1432.5 ± 133.7 1439.0 ± 83.2 1498.8 ± 190.8 1629.6 ± 45.8 1758.7 ± 189.2 b c cd d MximumHeight (H) (mm) 44.5 ± 0.1 46.2 ± 0.1 56.1 ± 0.2 61.8 ± 0.1 62.6 ± 0.5 64.0 ± 0.1 62.6 ± 0.2 67.0 ± 0.8 72.8 ± 0.2 74.7 ± 0.7 b cd d d At 100 min Trnsversl re (A) (mm 2 ) 921.5 ± 46.7 924.2 ± 12.9 1171.1 ± 47.7 1106.9 ± 74.0 1021.3 ± 142.1 1214.9 ± 74.8 c 1167.9 ± 85.9 999.6 ± 109.1 b 1107.3 ± 95.0 1129.4 ± 131.1 MximumHeight (H) (mm) 46.3 ± 0.1 40.9 ± 0.2 52.9 ± 0.2 cd 47.9 ± 0.1 d 46.7 ± 0.5 54.3 ± 0.1 d 51.7 ± 0.2 d 46.0 ± 0.5 b 51.6 ± 0.2 d 50.5 ± 0.5 d When the rtio between totl trnsversl re nd mximum height increse ( A/ H) t ech fermenttion time ws clculted, no stble evolution ws observed (Fig. 4). The instbility ws relted with the different velocity tht A nd H hd. In order to check this instbility, the correltion between the rc described by the lser light when it is projected onto the dough surfce (Fig. 4) nd the theoreticl rc ws used. As it is possible to observe in figure 4, n exmple of two doughs (left dough 2 nd right dough 9), in which the evolution of the rtio A/ H nd R 2 of Person re drwn, peks nd vlleys could be identified (between broken lines). Peks were considered when dt from R 2 chnge their derivtive sign from positive to negtive (derivtive cero vlue) nd the function vlue is equl or higher thn the previous pek. R 2 of Person nd A/ H hd n inverse tendency with time. When A/ H decresed, becuse the higher
velocity of H chnges, R 2 incresed, evolving the shpe of the dough surfce to theoreticl rc. Inverse behviour could be obtined when the A velocity ws higher. Figure 4. Evolution with fermenttion time (100 minutes) of the rtio between trnsversl re (A) nd the height (H) increse ( A/ H) (broken line) nd the rc correltion (continuous line) for dough number 9 (right) nd 2 (left). A B C D A Figure 5. Lser light projected onto the dough versus fermenttion time( A: 5 min, B: 30 min, C: 60 min, D : 100 min ) nd the theoreticl rc (green line). The recount of the number of peks t 100 min (NP 100 ) (Tble 4) showed how this number is relted to dough evolution nd could be used to discern the finl behviour of doughs (Fig. 6). In doughs which did not hve n importnt vrition in their trnsversl re between 100 min nd their finl fermenttion time (first doughs), the number of peks did not incresed, reching their highest number of peks. On the other hnd, doughs which incresed their trnsversl re lso incresed their number of peks (Tble 4). The filure of the gluten-network films seprting some bubbles hs been implicted in the correct gs retention nd therefore in the instbility of the fom structure of bred (Gn et l., 1995), which could produce vritions in the dough s evolution with time. Therefore, this could be the reson tht reorgniztion or non-uniform distribution of bubbles could led to totl or prtil collpse of the dough (Wng et l., 2011), mking chnges in the shpe of doughs which could be relted to the peks nd vlleys. According to this result, employing SL technique it is possible to chrcterize the
whet flours ccording to their fermenttion cpcity. Tble 4. Number of peks t 100 min (NP 100 ) nd t finl fermenttion time (NP f ) Dough NP 100 NP f de 1 7.5 ± 0 8 ± 1 NP100 2 8.5 ± 1.4 3 7 ± 0.7 4 6 ± 0.7 5 6 ± 0.7 6 6 ± 0 7 6 ± 0.7 8 4.5 ± 0.7 9 5 ± 0 10 4.5 ± 0.7 9 8.5 8 7.5 7 6.5 6 5.5 5 4.5 1 2 3 e cd b 8.5 ± 1 7 ± 0 7.5 ± 1 8 ± 1 7.5 ± 1 9.5 ± 1 7 ± 1 8.5 ± 1 9 ± 1 4 1000 1200 1400 1600 1800 Figure 6. Rtio between the trnsversl re (A 100 ) nd the number of peks t 100 minutes (NP 100 ) relted to the trnsversl re t the end of ech fermenttion time (A f ) for the ten doughs tested. 4. Conclusion The study demonstrted tht the Structured Light method (SL) cn provide useful informtion bout the behviour of dough during fermenttion. The monitoring of the fermenttion of dough showed tht whet flours with similr moisture, protein, gluten, dry-gluten, flling number nd lveogrph prmeters hve different fermenttion cpcities. The peks nd vlleys tht tke plce during fermenttion time, when the vrition of the totl trnsversl re is relted to the mximum height or with R 2 of Person (obtined when the curvture described by the lser light is djusted to theoreticl rc when it is projected onto the dough surfce), re directly relted with the fermenttion cpcity. A lower number of peks during 4 5 6 7 Af (mm2) 8 9 b b b b b b b b 10 the min fermenttion time (100 minutes) is relted to whet flours with high fermenttion cpcity. Hence the described technique could be used to check the fermenttion cpcity of whet flours ccording to their fermenttion behviour. 5. Acknowledgements We wish to thnk the Polytechnic University of Vlenci nd Generlitt Vlencin for the finncil support they provided through the PAID-05-011-2870 nd GVPRE/2008/170 Projects, respectively. 6. References Altenbch, S.B., Du Pont,F.M., Kothri, K.M., Chn, R., Johnson,E.L. & Lieu, D. (2003). Temperture, Wter nd Fertilizer Influence the Timing of Key Events During Grin Development in US Spring Whet. Journl of Cerel Science, 37, 9 20. Bjd, F. & Serš, I. (2011). Continuous monitoring of dough fermenttion nd bred bking by mgnetic resonnce microscopy. Mgnetic Resonnce Imging, 29(3), 434-442. Brk, S., Mudgil, D. & Khtkr, B.S. (2013). Reltionship of glidin nd glutenin proteins with dough rheology, flour psting nd bred mking performnce of whet vrieties. Food Science nd Technology, 51, 211-217. Brrer, G. N., Pérez, G. T., Ribott, P. D. & León, A. E. (2007). Influence of dmged strch on cookie nd bredmking qulity. Europe Food Reserch Technology, 225, 1 7. Brrer, G.N., Bustos M.C, Iturrig, L, Flores, S.K., León, A.E. & Ribott P.D. (2013). Effect of dmged strch on the rheologicl properties of whet strch suspensions. Journl of Food Engineering, 116(1), 233-239. Birch Anj N., Petersen Mikel A., Arneborg N. & Hnsen Åse S. (2013). Influence of commercil bker's yests on bred rom profiles. Food Reserch Interntionl, 52, 160 166. Boycı I.H., Willims, P.C. & Köksel, H. (2004). A rpid method for the estimtion
of dmged strch in whet flours. Journl of Cerel Science, 39(1), 139-145. Clderón-Domínguez, G.J., Chnon-Pérez, A.L. Rmos-Cruz, A.I., López-Lr, A.D., Tlple-Vldivi, G.F. & Gutiérrez-López. (2008). Frctl nd imge nlysis of Mexicn sweet bred bubble distribution; influence of fermenttion nd mixing time. G.M. Cmpbell, M. Scnlon, L. Pyle, K. Nirnjn (Eds.), Bubbles in food 2: Novelty, helth nd luxury Americn. Assocition of Cerel Chemists, St. Pul, MN, USA. Cocchi, M.,, Corbellinib M, Foc,G., Lucisnoc, M., Pgnic, M. A., Lorenzo T. & Alessndro, U. (2005). Clssifiction of bred whet flours in different qulity ctegories by wvelet-bsed feture selection/clssifiction lgorithm on NIR spectr. Anlytic Chimic Act, 544, 100 107. Dniel, C. & Triboi, E. (2000). Effects of Temperture nd Nitrogen Nutrition on the Grin Composition of Winter Whet: Effects on Glidin Content nd Composition. Journl of Cerel Science, 32(1), 45-56. Dobrszczyk, B.J. & Morgenstern, M.P. (2003). Rheology nd the bredmking process. Journl of Cerel Science, 38(3), 229-245. Fridi, H. & Fubion, J.M. (1990). Dough rheology nd bked product texture. Informtion Systems Division, Ntionl Agriculturl Librry (United Sttes of Americ). Gn, Z., Ellis, P.R. & Schofield, J.D. (1995). Gs Cell Stbilistion nd Gs Retention in Whet Bred Dough. Journl of Cerel Science, 21(3), 215-230. Jekle, M. & Becker, T. (2011). Dough microstructure: Novel nlysis by quntifiction using confocl lser scnning microscopy. Food Reserch Interntionl, 44(4), 984-991. Keskin S. O..Sumnu G.. Shi S. (2004). Bred bking in hlogen lmp microwve combintion oven. Food Reserch Interntionl, 37, 489 49. Lssoued N, Bbin P, Dell Vlle G, Devux MF, Reguerre AL. (2007). Grnulometry of bred crumb grin: contributions of 2D nd 3D imge nlysis t different scle. Food Reserch Interntionl., 40, 1087 97. Mills, E. N. C., Wilde, P. J., L. Slt, J. & Skeggs, P. (2003). Bubble formtion nd stbiliztion in bred dough. Food Bioproducts Process, 81, 189-193. Pour-Dmnb A.R. S, Jfry A. & Rfiee Sh. (2011). Monitoring the dynmic density of dough during fermenttion using digitl imging method. Journl of Food Engineering, 107(1), 8-13. Rdovnovic, N., Cloutier, S., Brown, D., Humphreys, D. G., & Lukow, O. M. (2002). Genetic vrince for gluten strength contributed by high moleculr weight glutenin proteins. Cerel Chemistry, 79(6), 843 849. Rizzello, C.G., Cod, R.,.Mzzcn, F., Minervini, D. & Gobbetti, M. (2012). Micronized by-products from debrnned durum whet nd sourdough fermenttionenhnced the nutritionl.texturl nd sensory fetures of bred. Food Reserch Interntionl, 46, 304 313. Updhyy, R., Debjni, G. & Mehr, A. (2012). Chrcteriztion of bred dough: Rheologicl properties nd microstructure Originl Reserch Article. Journl of Food Engineering, 109(1), 104-113. Verdú, S., Ivorr, E., J. Sánchez, A., Girón, J., Brt, J.M. & Gru, R. (2013). Comprison of TOF nd SL techniques for in-line mesurement of food item volume using niml nd vegetble tissues. Food Control, 33(1), 221-226. Wng, S., Austin, P. & Bell, S. It s mze: The pore structure of bred crumbs. Journl of Cerel Science, 54, 203-210.