Bruising pattern of table olives (‘Manzanilla’ and ‘Hojiblanca’ cultivars) caused by hand-held machine harvesting methods

Resea ch Pape B uising pa e n o able oli es (‘Manzanilla’ and ‘Hojiblanca’ cul i a s) caused by hand-held machine ha es ing me hods Ra ael R. Sola-Gui ado a,* , Se gio Bayano-Teje o b , Fe nando A agon-Rod iguez b , A aceli Pe~ na c , G ego io Blanco-Roldan b a Depa men o Mechanics, Uni e si y o Co doba, Campus de Rabanales, Co doba, Spain b Depa men o Ru al Enginee ing, Uni e si y o Co doba, Campus de Rabanales, Co doba, Spain c Resea ch Cen e CIAIMBITAL, Uni e si y o Alme ı´a, Alme ı´a, Spain a icle in o A icle his o y: Recei ed 24 May 2021 Recei ed in e ised o m 7 Janua y 2022 Accep ed 12 Janua y 2022 Published online 28 Janua y 2022 Keywo ds: B uise index Colou Damage F ui cha ac e isa ion Mechanical ha es ing Olea eu opaea This wo k p esen s a cha ac e isa ion o he ui and he b uising caused by some com- mon de achmen me hods (manual, s ick, shake comb, b anch shake ) and in e cep ion me hods (ne o padding) in common able oli e a ie ies. We ook pic u es o ui samples inside a special de ice, and he images we e p ocessed o ex ac cha ac e is ic pa ame e s o shape and size (numbe o spo s, Fe e diame e , ci cula i y, colou s …). Mo eo e , we s udied he ime e olu ion o b uising caused on he ui by a con olled impac . Finally, we de eloped a sys em ha allows synch onised o a ion o he ui wi h image cap u e o e alua e b uising on he whole olume o he ui . Ou esul s showed ha di e en ha es ing ea men s p oduced di e ences in he a e age numbe and diame e o spo s pe ui , as well as in he a e age a ea o he spo s pe ui o he di e en a ie ies. F ui colou o b uising can also se e as a con ol ac o o compu e ision cha ac e isa ion, o which eason we eco ded di e ences in he i mness o he b uised and non-b uised a eas o ui . The ha es ing me hod ha caused he highes median alues o b uise index was he shake comb, pa icula ly o ‘Manzanilla’ wi h an index o 1.59% on padding compa ed o 0.24% o ‘Hojiblanca’. Ne in e cep ion was also obse ed o inc ease he b uise index in ‘Manzanilla’ (5.85%). B uising assessmen ha only conside s a single pho og aph means ha a conside able amoun o b uising emains dis ega ded compa ed o he ac ual b uising on he whole olume o he ui . ©2022 The Au ho (s). Published by Else ie L d on behal o IAg E. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). *Co esponding au ho . E-mail add ess: [email p o ec ed] (R.R. Sola-Gui ado). A ailable online a www.sciencedi ec .com ScienceDi ec jou nal homepage: www.else ie .com/loca e/issn/15375110 biosys ems enginee ing 215 (2022) 188e202 h ps://doi.o g/10.1016/j.biosys emseng.2022.01.010 1537-5110/©2022 The Au ho (s). Published by Else ie L d on behal o IAg E. This is an open access a icle unde he CC BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/). 1. In oduc ion The cul i a ion o he oli e ee (Olea eu opaea L.) enjoys sus- ained g ow h a a global le el, cu en ly eaching 10.5 Mha (FAO, 2019). Sligh ly less han hal o he o al p oduc ion, 2.9 million onnes, is des ined o able oli es. Spain plays a majo ole in he able oli e sec o , wi h a p oduc ion o 0.5 million onnes (IOC., 2020). The ‘Manzanilla’, ‘Go dal’ and ‘Hojiblanca’ cul i a s (Campus, Degi mencioglu, &Comunian, 2018) a e he mos common comme cial a ie ies due o hei good pulp/pi a io, and g een p ocessing is one o he mos com- mon me hods employed o able oli es. The Spanish s yle, which p edomina es in his me hod (S anchez-G omez, Ga cı´a- Ga cı´a, &Ga ido-Fe n andez, 2013;Royal Dec ee 679/2016), means ha any ui de ec s equi e special a en ion as hey in luence quali y and, he e o e, he p ice ui can a ain on he ma ke . Any spo s on he ui 's su ace will de e mine i s ca ego y o comme cialisa ion (Royal Dec ee 679/2016Royal Dec ee 679/2016). Such de ec s can be classi ied acco ding o hei o igin (Riquelme, Ba ei o, Ruiz-Al isen , &Vale o, 2008), and b uising cons i u es he cha ac e is ic ype o damage p oduced du ing ha es ing. Mechanical ha es ing conside ably inc eases he le el o damage o esh ui such as able oli es (Hussein, Fawole, & Opa a, 2020). The main objec i e o he manual ha es ing conduc ed in mos able oli e o cha ds is o educe damage o he ui . This causes a conside able loss o p o i abili y since manual ha es ing me hods compe e wi h he highly mech- anised sys ems employed in mode n oli e g owing o he new in ensi ied plan ing sys ems. Ad ances in his ield a e o ien ed owa ds comp ehensi e mechanisa ion using com- bined ha es ing machine y (Fe guson e al., 2010;Sola- Gui ado, Cas illo-Ruiz, Blanco-Roldan, Gonzalez-Sanchez, & Cas o-Ga cı´a, 2020) and pos -ha es sys ems (Rejano, Mon a~ no, Casado, S anchez, &De Cas o, 2010;Zipo i, Fishman, Zelas, Subbo in, &Dag, 2021). In many a ming con ex s, mechanisa ion is no a alid solu ion due o he exis ing o og aphy, he small size o he a m o a lack o e- sou ces, all o which make he in oduc ion o machine y di icul . The only al e na i e on hese a ms is ha es ing wi h semi-mechanised sys ems (Be na di e al., 2018) such as b anch shake s and shake combs, o adi ional me hods like bea ing o hand picking, in which he ui alls on o ne s o suspended sys ems. These semi-mechanised ha es ing sys- ems de ach he oli e in a di e en way, so i seems logical o hink hey will also p oduce a di e en damage pa e n. Simila ly, since damage is ela ed o he le el o ene gy hey ecei e (Jim enez-Jim enez, Cas o-Ga cı´a, Blanco-Rold an, Agu ¨e a-Vega, &Gil-Ribes, 2012), he e may be a ela ionship wi h he in e cep ion sys em, due o he co ela ion be ween damage and he heigh o he oli e d op (Sa acoglu, Uce , & Oza slan, 2011). Opa a and Pa ha e (2014) lis di e en p ocedu es o cha ac e ise damage in di e en esh ui . Non-des uc i e echniques such spec al imaging, nuclea echniques, he - mal imaging and ul asound imaging, among o he s, ha e a g ea po en ial o de e mine he in e nal p ope ies o ui and de ec b uising (S ella e al., 2015;Mohammed Raju, Janna , Wang-Hee, Changyeun, &Byoung-Kwan Cho;Du e al., 2020). Compu e ision echniques a e mo e limi ed in ha hey can only cha ac e ise ex e nal damage o esh ui . Howe e , o he cha ac e isa ion o g een oli es ha will be p ocessed Spanish s yle (Campus e al., 2018), isible spec al imaging may be su icien o pe o m b uise cha ac e isa ion wi h adi ional, low-cos echnologies. Su ace damage means any de ec in he exoca p o a ui , and may be asso- cia ed wi h damage o he inne mos laye s, so damage migh be isible and measu able (Li &Thomas, 2014). I is di icul o es ablish empi ical o mula o he quan i ica ion o b uising, which depends on di e en pa ame e s such as a ie y and ma u i y s a us among o he s. A p ope cha ac e isa ion should ake in o accoun he amoun , shape and colou o he damage. B uise a ea is pa icula ly sui able o he quan i a i e cha ac e isa ion o damage. I he ela ionship be ween he b uise a ea and he o e all ui a ea is conside ed (Jim enez- Jim enez, Cas o-Ga cı´a, Blanco-Rold an, Gonz alez-S anchez, & Gil-Ribes, 2013) i is possible o es ablish a b uise index (BI). This classi ica ion allows o a quan i a i e quali ica ion o he damage, wi hou es ablishing disc e e le els, as p oposed by Hadi, Ahmad, &Akande, 2009. I is impo an o ake special ca e o quan i y ui b uising o e he o al su ace a ea o he ui . (Co kidi, Balde as-Ruı´z, Taboada, Se ano-Ca e on, & Galindo, 2006). The ela ionship o b uise index wi h he mos commonly employed able oli e ha es ing sys ems would p o ide aluable in o ma ion o imp o e he quali y o mech- anisa ion. Likewise, ano he pa ame e o in e es could be he cha ac e isa ion o he shape o damage in oli es, in line wi h he conside a ions o o he au ho s ega ding b uise calcula- ion in o he ui s (Ki hawee, Pa ha ee a , S i ung uang, & Slaugh e , 2011;Mohammed e al., 2017;Van Zeeb oeck e al., 2007). Image analysis can also de e mine he colou and ipe- ness o oli es (Guzm an, Bae en, Pie na, &Ga cı´a-Mesa, 2015). O he s udies (Beyaz, Ma ı´nez Gila, G omez O ega, &G amez Ga cı´a, 2019) indica e he in e es o applying simila me hods o de e mine he colou o he ype o ui damage known as b owning in he segmen a ion o oli e damage. B uising is changeable and equi es he s udy o i s e olu ion om i s beginning (Jim enez-Jim enez, Cas o-Ga cı´a, Blanco-Rold an, Gonz alez-S anchez, &Gil-Ribes, 2013). To de e mine oli e b uising quan i a i ely and quali a- i ely, isible imaging sys ems wi h compu e echniques ha e shown g ea po en ial o delimi ing he b owned ui a eas a e ha es ing (Jim enez-Jim enez, Cas o-Ga cı´a, Blanco-Rold an, Gonz alez-S anchez, &Gil-Ribes, 2013;Beyaz, € Ozkaya, & _ Ic¸en, 2017;Ponce, Aquino, Millan, &Anduja , 2019). The aim o his a icle is o cha ac e ise he ui and he b uise ha occu in g een able oli es, speci ically he ‘Hojiblanca’ and ‘Manzanilla’ a ie ies, employing compu e ision echniques a e he use o common ha es ing me hods. These pa ame e s we e e alua ed when ex e nal damage had s abilised and, in addi ion, du ing empo al e olu ion. The compa ison o damage quan i ica ion also ook in o accoun he en i e su ace o he ui e sus only one side o he ui . biosys ems enginee ing 215 (2022) 188e202 189 2. Ma e ials and me hods We conduc ed a se ies o ials, which consis ed o ha es ing oli es acco ding o di e en ea men s and analysing he pa e n o ex e nal damage caused o he ui s by means o image analysis. 2.1. Ob aining images o cha ac e ise he oli es The oli e samples we e placed in 0.18 m squa e polylac ic acid (PLA) ays, co e ed in blue oam wi h ellipsoid pe o a ions o he placemen o 20 ui s (Fig. 1). The samples we e placed in a closed ligh ing de ice (Sola-Gui ado, Bayano-Teje o, e al., 2020) wi h con olled ligh ing 13.95 ±0.57 lux, and a colou empe a u e o 5500 K om 4 LED ba s on he loo a 30, wi h di used ligh ing. A digi al came a (Nikon, D80, Tokyo, Japan) was placed o e o he samples, 570 mm abo e he ay su - ace, o ake pho og aphs a an ape u e se ing o /6.3, shu e speed E:1/125 s, ligh sensi i i y ISO:160, ocal leng h FL: 35 mm, and exposu e compensa ion EB:-1 EV. The eso- lu ion o each pho og aph was 3872 2592 pixels pe ay a ea wi h a ui esolu ion o 226 197 pixels. Each pho og aph was digi ally p ocessed by an expe using ImageJ so wa e (Na ional Ins i u e o Men al Heal h) o cha ac e ise he ui and i s b uised a ea. To do his, he pic u e was i s cu o ob ain 20 indi idual pho og aphs o each ui , iden i ied acco ding o i s ea men . Segmen a ion con e ed each pho og aph in o a bina y image o ex ac he ui measu emen s. Then, by adding he bina y pho o o he o iginal ui pho o, he backg ound was emo ed, lea ing only he ui . Finally, he b uised a ea o each ui was manually segmen ed o ob ain ano he pho og aph wi h he b uising a eas. Analysis o he pic u es epo ed he ollowing use ul pa ame e s o cha ac e ising he ui and b uising pa e n: - F ui and b uise a ea (mm 2 ): he a e age numbe o pixels o each ui o b uise scaled wi h i s pixel- o- millime e con e sion. - B uise index (%), BI: a io be ween he o al b uise a ea o each ui and i s ui a ea. - F ui leng h and diame e (mm). La ges and smalles dis ance be ween wo poin s on he ui a ea, usually coinciding wi h he majo and mino diame e s o he ellip ical shape o he ui . - Maximum Fe e diame e (mm) o he b uise spo : longes dis ance be ween any wo poin s along he spo selec ion bounda y. - Ci cula i y (#): he oundness o simila i y o a pe ec ci cle o he ui o b uised spo , which a ies om 0 o 1, wi h 0 co esponding o an in ini ely elonga ed poly- gon and 1 o a pe ec ci cle. - Numbe o spo s: numbe o b uised spo s pe ui . - Colou : coo dina es on he a e age RGB colou space o he pixels o each ui wi hou aking in o accoun he b uised a ea, o o he b uised spo i sel . 2.2. Ha es ing ea men s o ees and oli e samples We conduc ed he ials on wo di e en plo s o ‘Manzanilla’ able oli es and ano he wo plo s o ‘Hojiblanca’ des ined o g een p ocessing, loca ed in Co doba (Spain) du ing he mon hs o Sep embe and Oc obe o he 2018/19 and 2019/20 ha es ing campaigns. The ees we e unde i iga ion and wi hou bio ic o abio ic s ess. The ha es ing sys ems used o de ach he ui we e hose habi ually employed o his pu pose: manual picking (M) as a e e ence ea men , manual bea ing wi h a long ibe glass pole (B), semi-mechanised wi h a b anch shake (BS) (S ihl, SP 481, Waiblingen, Ge many), and semi-mechanised wi h a shake comb (SC) (Pellenc, P230, Pe uis, F ance). The ib a ion signals p oduced by he ha - es ing me hods on he main b anches we e eco ded using a MEMS iaxial accele ome e (Gul Coas Da a Concep s LLC X200-4, Wa eland, MS), wi h a measu emen ange o ±2000 ms 2 , a sensi i i y o 0.06 m s 2 and a sampling equency o 400 Hz. Table 1 summa ises he ib a ion pa e ns ha cha - ac e ised hese sys ems on he b anches. Each ee was ha - es ed om b anches loca ed be ween 1.5 and 2.4 m abo e he Fig. 1 eT ay wi h impac ed ui and sequence o impac caused by ee- all o he ball. biosys ems enginee ing 215 (2022) 188e202190 g ound by a a m wo ke , using he usual echnique employed on he plo . An ag icul u al ex ile ne ing (N), as usually used in oli e ha es ing, was placed on he loo , along wi h a padding su ace (P), consis ing o wo laye s o 3 mm hick poly inyl chlo ide illed wi h ai wi h a diame e o 1 m a a heigh o 0.1 m. Oli e samples we e aken om he padding o s udy he exclusi e incidence o he de achmen me hod. In he case o he ‘Manzanilla’, a ie y, which has g ea e suscep ibili y o b uising (Jim enez-Jim enez, Cas o-Ga cı´a, Blanco-Rold an, Gonz alez-S anchez, &Gil-Ribes, 2013), sam- ples we e also aken om he ne in a eas a om he padding o s udy he incidence o he in e cep ion means on his a ie y. In addi ion, andomly hand-collec ed samples wi hou ex e nal damage we e ha es ed and an impac o hi (H) was applied o he cen e o he ui . To pe o m he impac , a de ice was used in ee all om a heigh o 0.125 m wi h a s eel ball ha had a mass o 0.035 kg, applying an en- e gy o 0.043 J (Fig. 1). The ollowing ea men s we e he eby ob ained: MP, BS-P, SC-P, BeP o ‘Hojiblanca’ and MN, MP, BS-N, BS-P, SC-N, SC-P, BeN, BeP o ‘Manzanilla’ and H o bo h a i- e ies. The samples ob ained in each o he ea men s, we e kep d y a an a e age empe a u e o 23 C and pho o- g aphed 150 min a e ha es ing, when he b uise had s abilised (Jim enez-Jim enez, Cas o-Ga cı´a, Blanco-Rold an, Gonz alez-S anchez, &Gil-Ribes, 2013)., a oiding hose ha ci cumsc ibe he pe ime e o he plo . Th ee ows o ees we e andomly selec ed and, in each ow, h ee ees we e again andomly selec ed o apply he ea men s (1 ee exclusi ely used o 1 ha es ing me hod). When selec ing ows, hose co esponding o he pe ime e o he plo we e disca ded, as we e he ees a he beginning and end o he ow. Twen y ui samples (1 ay) we e aken om each ee o each ea men (BSeP, SC-P, BeP, BS-N, SC-N, BeN), esul ing in a o al 60 samples pe ea men and day. T ea men s wi h a common ha es ing me hod we e applied on he same ee, and samples we e ha es ed simul a- neously using he ne o padding su ace. Fo he e e ence ea men (M), oli es we e andomly ha es ed om he 3 ees in he ow be o e applica ion o he ea men s. In case o impac ea men (H), I oli e samples we e also collec ed om he 3 ees o he ea men s in he ow, on di e en a eas o he ee and wi hou p e ious b uising. These ials we e epea ed 4 imes pe campaign and plo , spaced app oxima ely 5 days apa , lea ing a minimum di e ence o 4 days in he e en o ain all. 2.3. Cha ac e isa ion o oli es and hei ex e nal damage We conduc ed di e en ypes o s udies o de e mine use ul pa ame e s o modelling oli es and hei b uising: 1. Size and shape cha ac e isa ion: Se e al geome ical pa ame e s we e aken o he ui (leng h, diame e , and ci cula i y), o he b uised spo s hey had (numbe o spo s, Fe e diame e and ci cula i y) and o all ea men s, ex ac ing he in o ma ion om image analysis. In addi ion, he ui mass (g) was measu ed wi h a digi al scale (G am, EH-500, Spain). 2. Fi mness assessmen : The pene a ion o ce needed o b eak he ui su ace was measu ed wi h a pene- ome e using a cylind ical 3 mm long and 2.4 mm diame e ip (IMADA Inc., DS2-11, USA) in non-b uised and b uised a eas ( he la e om spo a eas g ea e han 16 mm 2 ). 3. Colou de e mina ion: The colo ime ic cha ac e is ics o he oli e in ui a eas wi h damage (spo a ea g ea e han 16 mm 2 ) and wi hou damage we e measu ed o all ea men s using a colo ime e (Konica Minol a, CR- 400, USA) calib a ed wi h a D65 2illumina ion measu ed in CIELAB colou space. Simila ly, we de e - mined he RGB colou coo dina es o he images p ocessed. 4. B uise index and i s ime e olu ion: The b uise index was de e mined 150 min a e ha es ing o all ea - men s, using image analysis o ex ac he ui a ea and damage a ea. Mo eo e , images o he ui impac ed wi h he ee- all de ice (H) we e aken e e y 15 min up o 150 min, o e alua e he de elopmen o b uising o e ime. 5. Loca ion o he b uising on he su ace o he ui : Calcula ion o he b uise index no mally uses a single image, igno ing wha occu s in he unseen a eas, conside ing he ellipsoid geome y o he oli es. To e alua e ui b uising conside ing he en i e su ace o he ui , we designed a p o o ype (Fig. 2) o o a e he oli es on hei main axis. The de ice consis s o se e al gea s, wi h wo needles in hei cen es, whe e andomly selec ed ui samples a e placed. We e alu- a ed 60 ui samples pe ha es ing ea men and a- ie y, 150 min a e hei ha es . A mo o (4076 s eps pe u n) con olled by a mic ocon olle (A duino, Nano, I aly) o a ed he ui , shi ing 15a a ime o ake 24 pic u es pe ui in each posi ion o he e olu ion. To de e mine he amoun and loca ion o he b uising p oduced o e he whole su ace, he pic u es we e p ocessed o gi e a comple e wo-dimensional ep esen a ion o he oli e wi hou de o ma ion (Fig. 3): a. Each o he cap u ed images was c opped o a oid o e lapping o he same a ea, ci cumsc ibing an ellipse whose majo diame e coincided wi h he ui leng h Table 1 eVib a ion pa ame e s o ui -bea ing b anches wi h ha es ing sys ems. Ha es ing me hod Mean esul an accele a ion (m$s 2 ) F equency (Hz) S ick bea ing 560.7 ±214.3 a e Shake Comb 79.6 ±42.6 14.0 ±0.4 B anch shake 209.3 ±81.1 20.6 ±1.1 Values showed a e mean ±s anda d de ia ion. a Mean peak alue. biosys ems enginee ing 215 (2022) 188e202 191 and whose mino diame e is he ui pe ime e ðp∙diame e Þdi ided by 24. b. Each cu was joined o he consecu i e one. c. All pixels o he composi ion a e g ouped oge he o gi e con inui y o he composi ion while main aining he equa o ial line cons an . d. The emaining composi ion was enla ged by p ojec ion on o he cu a u e o he ellipse in he on iew, i.e. by making he dis ance be ween he peduncle and i s an- ipode coincide wi h hal he pe ime e o he ellipse de e mined in he on iew. In all cases, we in es iga ed he ela ionship ha exis s o all o he s udied a iables wi h bo h he ha es ing me hod and he in e cep ion me hod (in he case o ‘Manzanilla’). We also analysed he di e ences be ween he di e en a ie ies be o e inally s udying ela ionships exis ing wi hin he di e en s udy pa ame e s. 3. Resul s 3.1. F ui and ui b uising size and shape cha ac e isa ion Table 2 shows he cha ac e isa ion o shape and size o he ui s sampled o he wo ha es ing seasons acco ding o he di e en me hods. F ui mass, leng h and diame e a iables showed a no mal dis ibu ion (Kolmogo o eSmi no , p >0.05) o bo h a ie ies. A compa ison o he ha es ing me hods (ANOVA es , p >0.05; pos -hoc Tukey es , p >0.05) showed no signi ican di e ences in he a iables o mass, leng h and diame e o he wo a ie ies and seasons, no did we ind any signi ican di e ences be ween he ne and padding ca ching sys ems (ANOVA es , p >0.05; pos -hoc Tukey es , p >0.05). Howe e , we did ind signi ican di e ences (S uden 's - es , p<0.05) wi hin each a ie y in he compa ison o he 2019 and 2020 seasons: The diame e and leng h o ui o bo h a ie ies showed a s ong posi i e co ela ion wi h ui mass (‘Hoji- blanca’: Pea son Coe icien ¼0.88; Pea son Coe icien ¼0.89; p<0.05; ‘Manzanilla’: Pea son Coe icien ¼0.88; Pea son Coe icien ¼0.91; p <0.05). Ci cula i y did no show a no mal dis ibu ion (Kolmogo o eSmi no , p <0.05). Signi ican di - e ences we e ound in ci cula i y be ween a ie ies (Wilcox- oneManneWhi ney Tes , p <0.05), wi h a median alue and in e qua ile ange in ‘Hojiblanca’ o 0.865 (0.023) and 0.869 (0.026) in ‘Manzanilla’,wi h di use di e ences be ween ha - es ing me hods. Table 3 shows some ep esen a i e pa ame e s o he cha ac e is ic size and shape o he ui b uising spo s. The numbe o spo s pe ui did no ollow a no mal dis ibu ion. Fo bo h seasons and a ie ies, he numbe o spo s was signi ican ly di e en (K uskaleWallis, p <0.05) be ween he manual me hod and he o he de achmen me hods (pos -hoc Wilcoxon ank sum es wi h Holm adjus men , p <0.05), in which he e we e no di e ences be ween hem (pos -hoc Fig. 2 eDe ice de eloped o u ning oli es and aking images om di e en sides. biosys ems enginee ing 215 (2022) 188e202192 Wilcoxon ank sum es wi h Holm adjus men , p >0.05). In all cases, he numbe o spo s wi hin he same de achmen me hods inc eased in he ne ca ching compa ed o padding ca ching. The ci cula i y o spo s was signi ican ly di e en be ween de achmen me hods o ‘Manzanilla’ and ‘Hoji- blanca’ (K uskaleWallis, p <0.05; pos -hoc Wilcoxon ank sum es wi h Holm adjus men , p <0.05), and was lowe o he la e in all cases. The mean g ea e Fe e diame e o he spo was di e en (K uskaleWallis, p <0.05; pos -hoc Wilcoxon ank sum es wi h Holm adjus men , p <0.05) be ween a i- e ies and de achmen me hods, and sligh ly g ea e in ‘Hoji- blanca’, excep o he shake comb, whe e hey we e he same. 3.2. Fi mness assessmen Pene a ion o ce alues in he b uised and non-b uised ui zones ollowed a no mal dis ibu ion (Kolmogo o eSmi no , p>0.05) o bo h a ie ies. Signi ican di e ences (S uden 's - es , p <0.05) we e ound be ween he b uised and non- b uised ui zones in bo h a ie ies, wi h a alue o 14.02 ±1.92 N and 10.26 ±1.73 N o he b uised zone in ‘Hojiblanca’ and ‘Manzanilla’, espec i ely, and a alue o 17.11 ±1.73 N and 13.02 ±1.32 N o he non-b uised zones, in '‘Hojiblanca’ and ‘Manzanilla’, espec i ely. Wi hin each a- ie y he e we e also signi ican di e ences be ween he Fig. 3 eP ocess pe o med o he ep esen a ion o he en i e su ace o an oli e om he 24 pho os o a ui o a ed 15. Table 2 eF ui size and shape alues measu ed in wo ha es ing seasons acco ding o de achmen me hod wi h padded ca ching and oli e a ie y. De achmen me hod Va ie y Mass (g) a Leng h a Diame e a Ci cula i y b Manual ‘Hojiblanca’ 3.64 ±0.84 a 22.42 ±2.01 a 17.71 ±1.49 a 0.869 (0.024) a ‘Manzanilla’ 4.22 ±1.00 b 22.72 ±2.27 abc 18.62 ±1.86 b 0.877 (0.024) b B anch shake ‘Hojiblanca’ 3.58 ±0.83 a 22.43 ±2.02 a 17.60 ±1.56 a 0.864 (0.021) c ‘Manzanilla’ 3.91 ±0.98 c 22.92 ±2.11 bd 18.47 ±1.63 b 0.866 (0.026) ac Shake comb ‘Hojiblanca’ 3.56 ±0.70 a 22.36 ±1.64 a 17.68 ±1.35 a 0.863 (0.024) c ‘Manzanilla’ 3.90 ±0.97 c 22.59 ±2.13 ab 18.25 ±1.78 b 0.867 (0.025) a S ick bea ing ‘Hojiblanca’ 4.78 ±0.86 d 23.32 ±1.83 d 19.38 ±1.41 c 0.878 (0.021) b ‘Manzanilla’ 4.74 ±0.79 d 23.16 ±1.62 cd 19.35 ±1.34 c 0.879 (0.022) b Values shown a e mean ±s anda d de ia ion o he median and he in e qua ile ange in b acke s. Di e en le e s be ween ows o he same column indica e signi ican di e ences acco ding o. a No mal dis ibu ion (ANOVA, p <0.05; pos hoc pai wise wi h pooled s anda d de ia ion and Holm adjus men me hod, p <0.05) o . b Non-no mal dis ibu ion (K uskaleWallis, p <0.05; pos hoc Wilcoxon ank sum es wi h Holm adjus men me hod, p <0.05). biosys ems enginee ing 215 (2022) 188e202 193 b uised and non-b uised zones (pai ed - es , p <0.05), wi h an a e age educ ion in pene a ion o ce o 18% o ‘Hojiblanca’ and 21% o '‘Manzanilla’. In e ms o ha es ing me hods, he ui s o bo h a ie ies ollowed he same end in he non-b uised zones. The e we e signi ican di e ences o pene a ion o ce be ween he manual me hod, which had he highes alue, and he o he ha es ing me hods (ANOVA es , p <0.05; pos -hoc Tukey es , p <0.05). In he non-b uised zone, no di e ences we e ound be ween he b anch shake and shake comb o ‘Hojiblanca’ (S uden 's - es , p >0.05), wi h alues o 14.91 ±1.68 N and 14.10 ±1.94 N, espec i ely, whe eas di - e ences we e ound be ween hese me hods o '‘Manzanilla’ (S uden 's - es , p <0.05), wi h alues o 11.82 ±1.05 N and 9.73 ±1.61 N. A posi i e co ela ion o ui pene a ion o ce wi h ui weigh (Pea son coe icien p <0.05) was ound o bo h a ie ies in manual ha es ing me hods, conside ing ha he ui measu ed a e des ined o g een p ocessing. Pene a ion o ce alues in he non-impac ed zone o he ui showed a iabili y o e ime. Signi ican di e ences we e ound in bo h a ie ies be ween he i s ou weeks and he i h and six h weeks (ANOVA, pos -hoc Tukey p <0.05; K uskall-Wallis, pos -hoc WilcoxoneManneWhi ney Tes p<0.05). We obse ed ha he p opo ion o ui s wi h a highe ma u i y (highe Jaen index), inc eases as weeks go by (da a no shown). In addi ion, he alue o hei pene a ion o ce was di e en among ui s, excep o hose wi h indices o 0 and 1 (ANOVA, pos -hoc Tukey p <0.05; K uskall-Wallis, pos -hoc WilcoxoneManneWhi ney Tes p <0.05), which dec eased in alue as he ma u i y index inc eased. 3.3. F ui and ui b uising colou cha ac e isa ion Table 4 shows he componen s o CIELAB colou space measu ed wi h he colo ime e on he ui samples es ed. The alues o he ‘a’ componen showed no di e ences be- ween he di e en campaigns o each a ie y (Wilcox- oneManneWhi ney Tes , p >0.05), unlike he ‘L’ and ‘b’ componen s, which we e di e en o each a ie y (Wilcox- oneManneWhi ney Tes , p <0.05). In addi ion, we obse ed no signi ican di e ences be ween he di e en mechanised ha es ing me hods o be ween he padding o ne in e - cep ion me hods. The alues o all componen s we e highe in ‘Manzanilla’ han in ‘Hojiblanca’ in bo h seasons. In bo h a ie ies, nega i e ela ionships we e measu ed be ween he ‘a’ and ‘b’ componen o he CIELAB colou space (Spea man ho, p <0.05) wi h signi ican ela ionships ela ed o ui mass and diame e s (Pea son coe icien , p <0.05; Spea man ho, p <0.05) o ‘Manzanilla’. Al hough his s udy only e alua ed g een ui s, when conside ing di e en ma u i y indices, he analysis o he ui colou in RGB space ex ac ed om he pho og aphs shows a end o colou e olu ion om deep g een o black. (Fig. 4). F ui colou o e he six di e en weeks s udied was signi ican ly di e en (K uskall-Wallis, p<0.05) o each o he channels (RGB), wi h di e en ed and g een channel alues o each week (pos -hoc Wilcox- oneManneWhi ney Tes wi h Holm co ec ion, p <0.05). 3.4. B uise index cha ac e isa ion B uise index 150 min a e ha es ing The b uise index did no ollow a no mal dis ibu ion (Kolmogo o eSmi no , p <0.05). Figu e 5 shows he median alues ob ained o he wo a ie ies and he ha es ing ea men s udied. The e we e signi ican di e ences (Wil- coxoneManneWhi ney Tes , p <0.05) o he wo seasons be ween ‘Hojiblanca’ and ‘Manzanilla’ wi h g ea e damage su e ed by ‘Manzanilla’. Conside ing exclusi ely de achmen me hod, i.e. compa ing padding ea men s, he e we e no signi ican di e ences be ween a ie ies using he manual me hod and manual bea ing wi h s ick (K uskaleWallis, p<0.05; pos -hoc Wilcoxon ank sum es wi h Holm adjus - men , p <0.05). Howe e , signi ican di e ences we e ound be ween a ie ies using he b anch shake (BI ¼0.03 o ‘Hojiblanca’ and BI ¼0.33 o ‘Manzanilla’) o he shake comb (BI ¼0.24 o ‘Hojiblanca’ and BI ¼1.59 o ‘Manzanilla’), wi h signi ican di e ences be ween bo h me hods o he same a ie y (pos -hoc Wilcoxon ank sum es wi h Holm adjus - men , p <0.05). Rega ding he in e cep me hod, i.e., compa ing he same de achmen me hod be ween padding and ne , we ound di e ences o all sys ems, which we e Table 3 eF ui b uised spo size and shape alues measu ed in wo ha es ing seasons acco ding o ha es ing me hod and oli e a ie y. De achmen me hod Ca ching me hod Va ie y Numbe o spo s Fe e Diame e (mm) Ci cula i y Manual Padded ‘Hojiblanca’ 0 (1) a 2.35 (2.10) a 0.709 (0.235) ab Padded ‘Manzanilla’ 0 (1) a 1.62 (2.20) b 0.788 (0.314) c Ne 2 (3) b 2.21 (2.01) ab 0.674 (0.233) ab B anch shake Padded ‘Hojiblanca’ 1 (2) c 1.53 (1.10) b 0.725 (0.205) a Padded ‘Manzanilla’ 2 (5) b 1.34 (0.83) c 0.787 (0.191) c Ne 3 (4) d 1.99 (1.42) b 0.704 (0.198) ab Shake comb Padded ‘Hojiblanca’ 1 (2) c 3.04 (2.60) d 0.628 (0.211) d Padded ‘Manzanilla’ 2 (3) b 2.79 (2.16) ad 0.677 (0.201) b Ne 3 (2) d 4.41 (2.37) e 0.580 (0.162) e S ick bea ing Padded ‘Hojiblanca’ 0 (2) ac 2.08 (2.61) ab 0.666 (0.379) abd Padded ‘Manzanilla’ 0 (3) ac 2.37 (2.18) abd 0.664 (0.328) abd Ne 2 (3) b 2.28 (2.34) abd 0.628 (0.305) abd The alues ep esen ed a e median and in e qua ile ange. Di e en le e s indica e signi ican di e ences be ween ows o he same column (K uskaleWallis, p <0.05; pos -hoc Wilcoxon ank sum es wi h Holm adjus men me hod, p <0.05). biosys ems enginee ing 215 (2022) 188e202194 signi ican ly highe wi h a ne in e cep (K uskaleWallis, p<0.05; pos -hoc Wilcoxon ank sum es wi h Holm adjus - men , p <0.05). The highes alues we e ound using he shake comb o bo h seasons and a ie ies wi h g ea e alues in ‘Manzanilla’ (BI ¼1.59 o padding and BI ¼5.85 o ne ). I is possible o make a nominal classi ica ion by placing b uise index limi s (BI ¼0, 0<BI <1, 1<BI <3, 3<BI <5, BI >5) in o de o compa e he le el o damage s udied by o he au- ho s (Cas o-Ga cia, Cas illo-Ruiz, JimenezeJimenez, Gil- Ribes, &Blanco-Roldan, 2015: no damage, sligh damage, mode a e damage, se e e damage, mu ila ed). Howe e , his is only a p oposal o h esholds since hese au ho s used a isual classi ica ion so did no quan i y he a ea o b uising used o de ine he limi s. Figu e 6 shows he dis ibu ion o b uising in he sampled ui wi h ou p oposed classi ica ion. Wi h he h esholds, we ob ained a dis ibu ion o he le el o damage caused by each ha es ing me hod s udied (Fig. 7). B uise index and i s de elopmen o e ime Figu e 8 shows he empo al e olu ion o he b uise index om he ime he con olled me hod (H) impac ed he oli e. B uise index alues a di e en imes pos impac showed signi ican di e ences (S uden 's - es , p <0.05; Wilcox- oneManneWhi ney Tes , p <0.05) be ween a ie ies. The median b uise index alues we e i ed o a loga i hmic Table 4 eCIELAB colou space measu ed wi h he colo ime e on ui samples impac ed wi h he con olled ene gy me hod (H) and hose ha es ed wi h he shake comb, b anch shake and bea ing me hods using padding in he in e cep ion. De achmen me hod Va ie y L A b Con olled ene gy (H) (damage zone) ‘Hojiblanca’ 52.9 ±2.9 c 15.1 ±1.7 b 32.4 ±3.2 c ‘Manzanilla’ 52.6 ±3.8 c 11.5 ±2.9 c 30.3 ±3.8 d Con olled ene gy (H) (undamaged zone) ‘Hojiblanca’ 58.6 ±3.0 a 19.2 ±1.1 a 38.6 ±2.3 a ‘Manzanilla’ 62.1 ±2.5 b 18.8 ±3.2 a 40.8 ±3.5 b Manual ‘Hojiblanca’ 59.5 ±3.0 a 18.6 ±2.6 ad 37.8 ±2.8 a ‘Manzanilla’ 62.4 ±2.3 b 18.5 ±2.0 d 39.4 ±3.0 e Mechanical ha es ing (SC, BS, B) ‘Hojiblanca’ 59.0 ±3.1 a 19.0 ±1.3 a 36.5 ±4.6 ‘Manzanilla’ 61.9 ±4.5 b 18.3 ±4.0 d 38.4 ±2.9 a Values shown a e mean ±s anda d de ia ion. Di e en le e indica es signi ican di e ences be ween ows o he same column (K us- kaleWallis, p <0.05; pos -hoc Wilcoxon ank sum es wi h Holm adjus men me hod, p <0.05). *The colo ime e was placed on a andom a ea o he ui ega dless o whe he he a ea was damaged o no , excep o he con olled ene gy me hod, in which i was posi ioned jus abo e he a ea impac ed by he ball. Fig. 4 eColou e olu ion (R) du ing able oli e ipeness h oughou he ha es ing season (median o he alues in RGB space o Hojiblanca y Manzanilla). (Fo in e p e a ion o he e e ences o colo in his igu e legend, he eade is e e ed o he Web e sion o his a icle.) Fig. 5 eB uise index o each ea men o ui de achmen (M: manual by hand, BS: b anch shake , SC: shake comb, B: manual bea ing) and in e cep ion (N: ne s, P: padding) and o each oli e a ie y. The alues shown ep esen median and in e qua ile ange (K uskaleWallis, p <0.05; pos -hoc Wilcoxon ank sum es wi h Holm adjus men me hod, p <0.05). biosys ems enginee ing 215 (2022) 188e202 195 unc ion in ela ion o he ime pos impac . Figu e 9 illus a es he colou di e ences o he b uised spo s be ween he RGB colou channels o e ime. Colou di e ences be ween he wo a ie ies o ui o e he ime we e ound in he ed and g een channel (F iedman Tes p <0.05) bu no in he blue channel. The g ea es changes o e ime occu ed in he g een channel, whe e alues dec eased wi h a g ea e slope in ‘Manzanilla’. Loca ion o he b uising on he su ace o he ui Figu e 10 shows he di e ences in he b uise index be- ween analysing a single image ha pa ially cap u es he su ace o he ui , and he p oposed sys em, which cap u es he en i e ui su ace. The me hods based on b anch shake s and manual ha es ing showed signi ican di e ences be- ween he b uise index de e mined om a single pho o o om he whole pe spec i e in ue magni ude. Con e sely, ha es ing me hods wi h shake comb and bea ing showed no signi ican di e ences. In gene al, when he image o he sp ead ex e nal su ace o he ui is conside ed, he e is a ma ked inc ease in he pe cen age o b uising, a median 140%, conside ing ha 100% is he es ima ed b uising om a single unp ocessed zeni hal pic u e, al hough a high de ia ion exis s in some cases. Fig. 6 eDis ibu ion o b uising alues, bounda y lines o each damage ca ego y shown wi h e ical do ed lines. Fig. 7 eF equency o damage ca ego y o each ea men o ui de achmen (M: manual by hand, BS: b anch shake , SC: shake comb, B: manual bea ing) and in e cep ion (N: ne s, P: padding) and o each oli e a ie y. Di e en le e indica es signi ican di e ences be ween ea men s in he same damage ca ego y (pai wise wo-samples Z- es o p opo ions wi h Holm adjus men me hod, p <0.05). biosys ems enginee ing 215 (2022) 188e202196