Kiwifruit and apricot firmness measurement by the non-contact laser air-puff method

Transcription

1 Postharvest Biology and Technology 19 (2000) Kiwifruit and apricot firmness measurement by the non-contact laser air-puff method V. Andrew McGlone *, Robert B. Jordan Technology De elopment Group, HortResearch, Ruakura Research Centre, Pri ate Bag 3123, Hamilton, New Zealand Received 9 June 1999; accepted 21 December 1999 Abstract The laser air-puff method was investigated for non-destructive firmness measurements on kiwifruit (Actinidia deliciosa (A. Chev.) C.F. Liang et A.R. Ferguson cv. Hayward) and apricot (Prunus armeniaca L. cv. CluthaGold). The method involves delivering a sharp puff of air onto a stationary fruit whilst recording the resulting surface deformation with a laser displacement sensor. At a 65-kPa maximum puff pressure the deformations on both kiwifruit and apricots ranged from 0.1 mm for hard fruit to about 1 mm for very soft. A fruit stiffness value (E puff ) was calculated from the maximum deformation and compared with penetrometer firmness (F pen ) on cool-stored kiwifruit and mixed maturity apricots. Two measurements were made on opposite sides of each fruit using each technique and were averaged before regression analysis. Reasonable regression results (R , s=2.1 N) were obtained between F pen and E puff on kiwifruit over the penetrometer firmness range 0 30 N. The relationship showed a trend of increasing scatter with increasing kiwifruit firmness. With apricots, over a greater penetrometer range (0 80 N), the relationship between F pen and E puff was best described by a simple power law regression (R , s=8.2 N). Both sets of fruit results suggest the laser air-puff method is only suitable for coarse screening of fruit into two penetrometer firmness classes. Improved correlations may be possible by using higher puff pressures and/or more measurements per fruit Elsevier Science B.V. All rights reserved. Keywords: Firmness; Stiffness; Non-destructive; Penetrometer; Kiwifruit; Apricots 1. Introduction The laser air-puff method, invented by researchers at the University of Georgia (Prussia et al., 1993, 1994), is attractive in its potential as a high speed grading technique as no mechanical * Corresponding author. Tel.: ; fax: address: amcglone@hort.cri.nz (V.A. McGlone) contact is required between the fruit and the sensor. The method works by delivering a puff of air to the fruit and simultaneously measuring the surface deformation that occurs. The underlying theory is that the ratio of deformation to applied force can be linked to fruit stiffness and, further, that stiffness is related to the penetrometer firmness. Hung et al. (1999) recently reported on the development and use of a laser air-puff instru /00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved. PII: S (00)

2 48 V.A. McGlone, R.B. Jordan / Posthar est Biology and Technology 19 (2000) ment for measuring the firmness of peaches. They established that the instrument could provide a non-destructive firmness measure that was significantly correlated (R ) with the penetrometer firmness of peaches over a large range (0 120 N). We also have recently reported on the development of a similar laser air-puff instrument (Mc- Glone et al., 1999). That report concentrated on characterising the instrument and providing evidence that the method could indeed measure fruit stiffness. Preliminary kiwifruit measurements were reported that were encouraging in relating fruit stiffness to penetrometer firmness (R ) over a narrow penetrometer firmness range (0 12 N). The objective of the current research was to investigate the laser air-puff method for fruit firmness measurement on kiwifruit and apricots. For kiwifruit, the primary interest is in sorting fruit after long-term cool storage when consignment averages go well below 30 N and the presence of soft fruit puts consignments in jeopardy in terms Fig. 1. A schematic diagram (not to scale) of the laser air-puff instrument. The nozzle is shown in cross-section with a shaded line following the vertical laser beam path to the fruit and then the imaging line (off-angle) back to the sensor. of industry regulations and final consumer acceptance in the market place. Consumers will generally not accept fruit 6 N (Stec et al., 1989) and this results in setting thresholds back at the cool store at anywhere between 8 and 15 N depending on the exact market requirements and post-storage handling conditions. With apricots, the interest is in harvest-time sorting of mixed maturity consignments to remove immature or over-mature fruit. Penetrometer firmness has been suggested as a standard for determining maturity in stonefruit, and in particular, over-maturity (Crisosto, 1994). 2. Materials and methods 2.1. The laser air-puff instrument The major components of the laser air-puff instrument were a tank of pressurised air, a solenoid switching valve, an outlet nozzle and a laser displacement sensor (Fig. 1). Fruit to be tested were positioned directly below the nozzle and supported, when necessary, in a cradle made from soft modelling clay. The cradle prevented sideways movements of the fruit under the puff and also minimised deformation on the bottom side of the fruit by spreading the load over a large contact area. The solenoid valve, nozzle and displacement sensor were all fixed in position on a shifting arm that was adjusted, prior to most measurements, to give a nozzle-fruit distance of 23.3 mm that brought the displacement sensor reading close to its zero setting. A short blast of air, the puff, was delivered to the fruit by computer control of the open duration (typically 80 ms) of the solenoid valve (VS3145, SMC, USA). The tank supplying the air had a capacity of 50 l and was regulated to a typical set air pressure of 100 kpa. The nozzle had an internal diameter of 8.2 mm and this delivered a steeply peaked pressure profile at the fruit surface of diameter, at half-width, of 10 mm. Fruit surface deformations during a puff were measured with the laser displacement sensor (Keyance LB-081, Osaka, Japan), which had a range of 15 mm around the zero setting and a resolution of 8 m. The laser beam of the sensor passed through a glass

3 V.A. McGlone, R.B. Jordan / Posthar est Biology and Technology 19 (2000) Fig. 2. A deformation curve (solid line) and a repeat (dotted line) measured at the same site on a soft apricot (F pen 5 N). The peak deformation was measured as shown, for the first curve, and was 0.54 mm for both curves. Puff duration was 80 ms and the tank pressure was 100 kpa. Timing of solenoid control pulse shown at top. window on top of the nozzle holder, down through the nozzle, and then onto the fruit (Fig. 1). The sensing element of the sensor was on the outside of the nozzle holder and had an unobstructed view of the laser spot on the fruit. Bringing the laser beam of the sensor directly down the nozzle constitutes the main difference between our instrument and the University of Georgia instruments, which have either the laser beam (Fan et al., 1994) or the air-puff (Hung et al., 1999) striking the fruit surface at a non-normal angle. Whilst we certainly enjoy the technical elegance of our particular solution, it will not matter to the general problem of fruit firmness measurement which arrangement is employed. The raw data from the instrument was called the deformation curve and consisted of a timed sequence of displacement sensor readings, recorded at a sampling rate of 1 khz, during the application of a puff to the fruit. A typical deformation curve (Fig. 2) shows four distinct stages. The first is up to 150 ms of baseline displacement readings of the undeformed fruit surface. The solenoid signal was delayed by 100 ms to ensure a reasonable baseline period. The shifting arm of the instrument had been previously adjusted to give a baseline reading close to zero. At 150 ms, some 50 ms after initiation of the solenoid signal, the second stage occurs with displacement readings climbing quickly due to the puff impacting on the apricot. The readings level off (the third stage) after a further 50 ms and reach a maximum after about 220 ms total elapsed time. The final stage occurs after 260 ms elapsed time, and some 70 ms after cessation of the solenoid signal, as the displacement readings decline exponentially back towards the baseline reading in response to diminution of the puff. The deformation curves are not necessarily exactly repeatable and, in particular, the baseline reading is not recovered during the final stage (Fig. 2). Softer fruit, or higher peak puff pressure, generally resulted in a greater baseline discrepancy. Deformation curves for elastic objects (e.g. rubber balls) always showed complete recovery and so the baseline discrepancy observed with fruit is due to non-recoverable strains occurring at lower stress limits than those applying to elastic objects. These non-recoverable strains never revealed themselves in visible damage on the fruit and did not affect the repeatability, at least for successive pairs of measurements, of the firmness parameter extracted from the deformation curves (Fig. 2). Hence it appears that the fruit are not damaged by the puff although no rigorous damage or repeatability studies have yet been done to confirm this. The key firmness parameter extracted from the deformation curve was the peak deformation, D, calculated as the difference between the maximum displacement reading and the average baseline reading occurring during the first stage (Fig. 2). Peak deformation is inversely related to fruit firmness with a larger deformation corresponding to a softer fruit. Peak deformation measurements are repeatable for a range of fruit and for fruit firmness at tank pressures up to 200 kpa (McGlone et al., 1999). Fruit stiffness, defined here as E puff (the modulus of elasticity by the laser puff method), was calculated from the peak deformation measurement, D, using a formula based on the Boussinesq theory of die loading (Mohsenin, 1986) E puff =P (1 2 ) /2D (1)

4 50 V.A. McGlone, R.B. Jordan / Posthar est Biology and Technology 19 (2000) where P is the peak puff pressure on the fruit, is the poisson ratio and is an equivalent die radius of the puff. The peak puff pressure P was calculated from the tank pressure P tank with an empirically derived formula (McGlone et al., 1999). A poisson ratio of =0.4 was used in all calculations and was chosen (not measured) on the basis of being a mid-range value for fruit between the apparent extremes of =0.3 for apples and = 0.5 for potatoes (Mohsenin, 1986). The equivalent die radius has been estimated at 2 mm for the measurement conditions in this study. Justification for the use of the Boussinesq formula and the estimate of =2 mm is given in McGlone et al. (1999) Kiwifruit data set Kiwifruit ( Hayward ) from four orchard lines were obtained from a commercial cool store and place in a controlled laboratory cool store operating at a set temperature of 0 C. The fruit were all between 0 and 4 weeks post-harvest, in a single weight band of g and 20 N penetrometer firmness at procurement. At intervals over a period of 24 weeks the fruit were sampled from the laboratory cool store and warmed to room temperature ( 20 C) overnight before being measured for firmness, first with the laser air-puff instrument and then the penetrometer. At each sampling, between 15 and 20 fruit were randomly chosen from each of the four orchard lines. In total 379 kiwifruit were sampled over the 24-week period. Two peak deformation measurements, at 65 kpa peak puff pressure and 80 ms puff duration, were made on opposite sides of each fruit. The deformation measurements were each transformed into E puff measurements before being averaged to provide a mean value for the fruit. Penetrometer measurements were made at the same two sites on each kiwifruit. Data analysis centred on establishing the strength of relationship between the mean E puff and F pen measurements made on each fruit. In addition, the measurement pairs per fruit, for both E puff and F pen separately, were compared by regression analysis to assess the degree of within-fruit variability associated with each measurement method Apricot data set One orchard line of 140 CluthaGold apricots was chosen for this study. The fruit were harvested within a 2-h picking session and included a wide range of maturity. The fruit were immediately sorted by eye into seven colour grades, consisting of 20 fruit each, and were sent overnight to the research laboratory where the laser puff and penetrometer firmness measurements were made. Central sites on each opposing cheek of the fruit were chosen to provide two firmness measurements per fruit. The firmness measurement procedures and analysis methods were then exactly those of the kiwifruit data set Penetrometer firmness Penetrometer firmness (F pen ) measurements on kiwifruit and apricots were made in accordance with standard industry practice. Hand-held penetrometers (Effegi) were used with a 7.9-mm diameter plunger. Two measurements were made on opposite sides of the fruit (same sites as E puff measurements) and then averaged to give a mean value for the fruit. A 1-mm thick slice of skin was removed from each measurement site prior to the measurement. The maximum depth of penetration was 8 mm and, as is industry practice, the rate of penetration was subjectively controlled at about 5 mm/s (2 s to reach maximum depth). Penetrometer measurements are reported in Newtons (N) with 10 N being approximately 1 kgf (standard industry units of kilogram-force). 3. Results 3.1. Kiwifruit firmness The F pen distribution for the kiwifruit data set ranged from 2 to 29 N with a mean value of 9.7 N and standard deviation of 4.7 N. As fruit softened during storage the peak deformations measured with the laser air-puff method ranged from as low as 0.1 mm for hard fruit ( 20 N) to as high as 1 mm on very soft fruit ( 5 N). The relationship between F pen and E puff was strongly

5 V.A. McGlone, R.B. Jordan / Posthar est Biology and Technology 19 (2000) Fig. 3. Kiwifruit penetrometer firmness F pen against the laser air-puff stiffness E puff. The line is a linear regression fit to the data (R 2 =0.80). linear although the data look to be increasingly more scattered above 10 N penetrometer firmness (Fig. 3). Regression statistics of R 2 =0.80 and s (mean residual error) =2.1 N were obtained for F pen against E puff. The mean residual error is 22% of the mean fruit firmness (9.7 N). The E puff measurements made on opposites sides of the fruit, reflecting the within-fruit variability, are only moderately correlated to each other (Fig. 4(b)) with regression statistics of R 2 =0.66 and s=0.37 MPa. In contrast the F pen measurements Fig. 5. Apricot penetrometer firmness F pen against laser airpuff stiffness E puff. The line is a power law regression fit to the data (R 2 =0.80). from opposites sides are highly correlated (Fig. 4(a)) with regression statistics of R 2 =0.93 and s=1.3 N Apricot firmness The F pen measurements of the apricot data set ranged from 2 to 93 N with a mean of 22.8 N and a standard deviation of 18.5 N. Peak deformations ranged from 0.1 to 1 mm in going from very hard to very soft fruit, respectively. The plotting Fig. 4. Correlation scatter plots between penetrometer firmness (a) and laser air-puff stiffness (b) measurements made on opposites sites on each kiwifruit (R 2 =0.93 and 0.66, respectively). Straight lines are the 1:1 relationship.

6 52 V.A. McGlone, R.B. Jordan / Posthar est Biology and Technology 19 (2000) of F pen against E puff (Fig. 5) shows slight curvature, and a power law relationship F pen =2.6 (E puff ) 1.7 gave the best regression fit with an R 2 =0.8 and s=8.3 N. The mean residual error is equivalent to 36% at the mean fruit firmness of 23 N. The E puff measurements made on opposites sides of the fruit, reflecting the within-fruit variability, are only moderately correlated to each other (Fig. 6(b)) with regression statistics of R 2 = 0.64 and s=0.37. The corresponding F pen measurements from opposites sides are better correlated (Fig. 4(b); R 2 =0.8, s=8.3 N). 4. Discussion The reasonably high correlations (R 2 0.8) between F pen and E puff with the kiwifruit and apricot data sets means that the laser air-puff method can assess the same general firmness differences as the penetrometer method does with cool-stored kiwifruit and ripening apricots. For the kiwifruit data set the correlation is linear over the 2 30-N range of F pen measurements (Fig. 3). For the apricots the correlation was less linear and better modelled as a power law relationship over the F pen range 0 90 N (Fig. 5). The correlations are slightly better than those achieved by Hung et al. (1999) in using the laser air-puff method on peaches (R 2 0.8; value depending on cultivar). However a fair comparison is not possible because Hung et al. (1999) have used the peak deformation (D) directly, rather than a derived measure such as stiffness (E puff ), and re-expression as stiffness might well have improved the linearity and significance of their correlations. Stiffness is, in broad terms, the ratio of stress to strain in a loaded material and as such is a completely different physical property from the maximum stress or rupture force property that the penetrometer measures. Nevertheless, and as observed in this study, the empirical evidence (Abbott et al., 1997) is that stiffness and rupture force properties are often linearly correlated. In terms of penetrometer firmness prediction the results suggest only moderate accuracy and would probably only allow a coarse screening of fruit into high or low firmness grades. The absolute values of the mean residual errors, in regressing F pen against E puff, are quite different at 2.1 and 8.3 N for the kiwifruit and apricot data sets, respectively. However the data set ranges are also quite different ( 30 and 80 N, respectively) and in both cases the residual errors represent the same 45% fraction of the respective distributional standard deviations. Grading scenarios can be created, on the basis of the regression models for F pen against E puff,to get estimates of likely grading error rates. To reject, with at least 95% confidence, all kiwifruit Fig. 6. Correlation scatter plots between penetrometer firmness F pen (a) and laser air-puff stiffness E puff (b) measurements made on opposites sites of each apricot (R 2 =0. 80 and 0.64, respectively). Straight lines are the 1:1 relationship.

7 V.A. McGlone, R.B. Jordan / Posthar est Biology and Technology 19 (2000) with an actual F pen measurement below 10 N would require a sorting threshold on the laser air-puff predictions at about 14.2 N (twice the mean residual error above the 10-N maximum). The errors around this threshold can be estimated and result in an error rate of 23% of the fruit actually 10 N being rejected. With the kiwifruit data set there is also increasing scatter observable at higher firmness values (Fig. 3) which will make the grading error rates even greater for firmer fruit. The increasing scatter is most likely due to the inverse proportionality between E puff and the deformation D (Eq. (1)) that effectively amplifies errors in D, at lower deformations, relative to those at greater deformations. In addition, the fractional error in D becomes necessarily larger at lower deformation because of the finite resolution of the laser displacement sensor ( mm). For apricots a penetrometer firmness threshold at 20 N might be necessary to reject fruit that are too soft to survive days of delay in reaching distant retail markets. To be at least 95% confident of removing all such soft fruit, a sorting threshold at about 37 N is necessary on the laser air-puff predictions. This would mean that 26% of the fruit 20 N were being rejected in error. In all of this the laser puff method would appear at least comparable in accuracy to other non-destructive methods that have been recently reported. Kiwifruit studies with the impact method (McGlone and Schaare, 1998), the dynamic force-deformation method (Abbott and Massie, 1996) and sonic resonance method (Abbott and Massie, 1998) all reveal similar measurement errors of around 2 N for a 10-N penetrometer firmness. There appears to be no formally published work on the non-destructive firmness testing of apricots. For another stonefruit, peaches, a recent article (Stone et al., 1998) reports a linear correlation between the penetrometer and acoustic impulse methods of R 2 =0.76 where the measurements are averages made on opposite cheeks. That correlation is only slightly lower than that achieved here with the laser airpuff method. The prediction errors might be reduced by using higher puff pressures to improve the correlation between F pen and E puff. Correlations between the two properties depend on the measurement conditions, and in particular, how damaging or non-recoverable the strains are during the stiffness measurement. In general the larger the strains and the closer the stresses are to the maximum sustainable by the material then the higher the expectation of better correlations (Bourne, 1982). The need for the measurement to be non-destructive means the stress applied by the puff must be below the maximum sustainable by the fruit. The maximum pressure on the fruit during a puff (65 kpa) can be achieved with the 7.9-mm diameter penetrometer tip applied to the fruit with a force of 3.2 N. This force is significant considering the 5 N or less penetrometer firmness of the softer fruit. Therefore it is probably not realistic to increase the puff pressure to improve the correlations, without significantly damaging softer fruit. Another possibility for improving the correlations would be to take averages over more measurements per fruit, especially for the stiffness measurements that have much higher within-fruit variability than the penetrometer measurements (Figs. 4 and 6). Such an approach has been used with other non-destructive methods, involving averages over four or more measurement sites per fruit, to improve firmness predictions for peaches (Stone et al., 1998) and melons (Ozer et al., 1998). We speculate that the higher variability recorded here for E puff, compared to F pen, is most likely due to surface morphology differences between measurement sites that affects the laser air-puff method but not the penetrometer method for which the surface layer has been removed. However, it could also be that the fruit stiffness property is less homogenous around a fruit than the rupture force property that the penetrometer measures. Accurate stiffness measurements on excised tissue from around the fruit would be necessary to resolve this issue. The laser air-puff method should be suited to any sorting situations where fruit firmness is judged on the basis of stiffness alone. Stiffness measures of firmness are used in industry but are often quite subjective, such as in hand-sorting of produce, or slow as in the standard compression method (ASAE Standards, 1984). We are not aware of any sorting situations that enforce a

8 54 V.A. McGlone, R.B. Jordan / Posthar est Biology and Technology 19 (2000) stiffness type firmness measure, and readily agree with Abbott et al. (1997) that most fruit and vegetable industries appear to have accepted the conventional rupture force type methods, such as the penetrometer, as the standard firmness or maturity measure. A shift away from these standard measures, towards a non-destructive stiffness type measure, would require a large research effort to provide the industries with the necessary quality evidence (e.g. sensory trials) and market confidence in a new measure. To date, research initiatives along this path have only served to confirm that rupture force type measures provide the best standard methods (Pitts et al., 1997) despite their many known limitations (Abbott et al., 1997). In the case of the laser air-puff method further work might also be necessary to resolve the cause, and/or reduce by taking averages over more measurements per fruit, the high with-in fruit variability of the E puff measurements. 5. Conclusion The laser air-puff method provides fruit stiffness estimates that correlate reasonably highly with penetrometer firmness measurements on kiwifruit and apricots. However, the correlations against penetrometer firmness are not yet high enough to provide accurate predictions for more than a very coarse screening of fruit into two penetrometer firmness classes. So whilst the method continues to appeal as a high speed grading technique, because it is fast and non-contact, it is unlikely to find application in practical fruit firmness grading situations where accuracy is judged solely on the basis of penetrometer firmness. Options for improving the correlations include using a higher puff pressure and/or making more than two measurements per fruit. Acknowledgements We thank Barry Stevenson for mechanical design and construction of the laser air-puff instrument and Jill McLaren for organising the apricot. data set for us. This research was supported by the New Zealand Foundation for Research, Science and Technology. References Abbott, J.A., Massie, D.R., Nondestructive dynamic force/deformation measurement of kiwifruit firmness (Actinidia deliciosa). Trans. Am. Soc. Agric. Eng. 38, Abbott, J.A., Massie, D.R., Nondestructive sonic measurement of kiwifruit firmness. J. Am. Soc. Hort. Sci. 123, Abbott, J.A., Lu, R., Upchurch, B., Stroshine, R., Technologies for nondestructive quality evaluation. Hortic. Rev. 20, ASAE Standards, 31st ed., S Compression tests of food materials of convex shape. ASAE, St. Joseph, MI, pp Bourne, M.C., Food Texture and Viscosity: Concept and Measurement. Academic Press, New York. Crisosto, C.H., Stone fruit maturity indices: a descriptive review. Postharvest News Inf. 5, 65N 68N. Fan, S., Prussia, S.E., Hung, Y., Evaluating the UGA laser-puff food firmness detector. Am. Soc. Agric. Eng. Paper Hung, Y.-C., Prussia, S.E., Ezeike, G.O.I., Nondestructive firmness sensing using a laser air-puff detector. Postharvest Biol. Technol. 16, McGlone, V.A., Schaare, P.N., Design and performance of a fruit firmness grader. Acta Hortic. 464, McGlone, V.A., Ko, S.M.W., Jordan, R.B., Non-contact fruit firmness measurement by the laser air-puff method. Trans. Am. Soc. Agric. Eng. 42, Mohsenin, N.N., Physical Properties of Plant and Animal Materials. Gordan and Breach, New York. Ozer, N., Engel, B.A., Simon, J.E., A multiple impact approach for non-destructive measurement of fruit firmness and maturity. Trans. Am. Soc. Agric. Eng. 41, Pitts, M., Cavalieri, R., Drake, S., Fellman, J., Evaluating apple firmness sensors. Washington State Univ. Tree Fruit PostHarvest J. 8, Prussia, S.E., Hung, Y., Astelford, J. Jr., A non-contact firmness sensor. Am. Soc. Agric. Eng. Paper Prussia, S.E., Astelford, J.J., Hewlett, B., Hung, Y.-C., Non-destructive firmness measuring device. US Patent number , 13 December. Stec, M.G.H., Hodgson, J.A., MacRae, E.A., Triggs, C.M., Role of fruit firmness in the sensory evaluation of kiwifruit (Actinidia deliciosa cv Hayward). J. Sci. Food Agric. 47, Stone, M.L., Armstrong, P.R., Chen, D.D., Brusewitz, G.H., Maness, N.O., Peach firmness prediction by multiple location impulse testing. Trans. Am. Soc. Agric. Eng. 41,