Full Terms & Conditions of access and use can be found athttp://www.tandfonline.com/action/journalInformation?journalCode=ldrt20

Download by: [188.40.254.179] Date: 02 April 2017, At: 23:38

Drying TechnologyAn International Journal

ISSN: 0737-3937 (Print) 1532-2300 (Online) Journal homepage: http://www.tandfonline.com/loi/ldrt20

Improvement of kiwifruit drying using computervision system (CVS) and ALM clustering method

Mohammad Hossein Nadian, Mohammad Hossein Abbaspour‐Fard,Hassan Sadrnia, Mahmood Reza Golzarian, Mohammad Tabasizadeh & AlexMartynenko

To cite this article: Mohammad Hossein Nadian, Mohammad Hossein Abbaspour‐Fard, HassanSadrnia, Mahmood Reza Golzarian, Mohammad Tabasizadeh & Alex Martynenko (2017)Improvement of kiwifruit drying using computer vision system (CVS) and ALM clustering method,Drying Technology, 35:6, 709-723, DOI: 10.1080/07373937.2016.1208665

To link to this article: http://dx.doi.org/10.1080/07373937.2016.1208665

Accepted author version posted online: 19Aug 2016.Published online: 19 Aug 2016.

Submit your article to this journal

Article views: 73

View related articles

View Crossmark data

DRYING TECHNOLOGY 2017, VOL. 35, NO. 6, 709–723 http://dx.doi.org/10.1080/07373937.2016.1208665

Improvement of kiwifruit drying using computer vision system (CVS) and ALM clustering method Mohammad Hossein Nadiana, Mohammad Hossein Abbaspour‐Farda, Hassan Sadrniaa, Mahmood Reza Golzariana, Mohammad Tabasizadeha, and Alex Martynenkob

aDepartment of Biosystems Engineering, Faculty of Agriculture, Ferdowsi University of Mashhad, Mashhad, Iran; bDepartment of Engineering, Faculty of Agriculture, Dalhousie University, Truro, Nova Scotia, Canada

ABSTRACT The objective of this work was to improve kiwifruit drying with respect to quality. The drying of kiwifruit slices was studied with hot air drying (HAD) and hybrid hot air-infrared drying (HID) at 50, 60, and 70°C, air velocity of 0.5, 1.0, and 1.5 m/s and thickness of 3, 5, and 7 mm using full factorial design. Shrinkage and color changes of kiwifruit slices were monitored in real time with a computer vision system (CVS). The drying treatments were ranked based on desirability function, using the combination of minimal color change, minimal shrinkage, and drying time as decision criteria. The proper conditions of kiwifruit drying were determined by average linkage method clustering as: temperature, 70°C; thickness, 3 mm; air velocity, 1.5 m/s (HAD) or 0.5 m/s (HID). Finally, a hybrid approach with HID at the first stage and HAD at the second stage of drying was proposed as a reasonable compromise between drying time and product quality.

KEYWORDS Clustering; hot air; infrared; kiwifruit; online monitoring

Introduction

Drying is one of the most popular operations in the food, fruit, and vegetable processing industry. Various methods for fruit and vegetable drying, such as hot air,[1] infrared,[2] microwave,[3] freeze drying,[4]

ultrasound,[5] vacuum,[6] and hybrid drying[7] have been extensively used. Each method has its own advantages and disadvantages, which determine drying time, shrinkage, and quality degradation rates.

Decreasing of drying time is associated with more efficient use of energy and increase in profitability, which is the paramount for food industry.[8] One possible way to decrease drying time is infrared heating.[9] This method is especially suitable for thin-layer drying where material surface is largely exposed to infrared radiation.[10] How-ever, high-intensity infrared radiation can lead to increase in product temperature and scorching to the extent of fire and explosion.[11] The application of hybrid hot air- infrared drying (HID) could be a good alternative for heat-sensitive materials such as food products.[9] In general, hybrid drying technologies allow shorter drying time, leading in a favorable improvement in the energy required per unit of water removed.[12,13]

The product temperature is a crucial parameter affecting not only duration of the drying process but

also color changes and shrinkage.[14] Vega et al.[15]

reported that heat-sensitive products can be subjected to high air temperatures at the first stage of drying. Therefore, a good strategy of HID drying could be fast removal of surface moisture with infrared during the initial stage of drying, and control of product tempera-ture at the second stage to achieve the desirable balance between drying kinetics and product degradation as well as to ensure safe operation.

The appearance of dried materials is very important for consumers.[12,16,17] Color and shape are as the primary quality parameters, when it comes to consumer acceptance of dried products.[18] However, the drying conditions, which ensure minimum drying time do not usually correspond to the requirements of the highest quality.[19] Consequently, there is a need for improvement of the hybrid process settings, according to the demands on quality as well as drying-time criteria.

Multiple studies reported effect of drying conditions on drying kinetics and quality of kiwifruit. Maskan[20]

found that hot air drying (HAD) had little influence on color of kiwifruits as compared to microwave and hybrid hot air–microwave methods. Shrinkage of kiwifruits during microwave drying was greater than

CONTACT Mohammad Hossein Abbaspour‐Fard abaspour@um.ac.ir Department of Biosystems Engineering, Faculty of Agriculture, Ferdowsi University of Mashhad, Azadi Square, Mashhad 9177948974, Razavi Khorasan Province, Iran. Color versions of one or more of the figures in this article can be found online on at www.tandfonline.com/ldrt. © 2017 Taylor & Francis

HAD, but reduced with hot air–microwave drying.[21]

According to Orikasa et al.,[22] drying temperature and time of drying increased decomposition rate of the L-ascorbic acid (AsA) content in kiwifruit slices.

Computer vision systems (CVS) as noncontact/ destructive, reliable, and fast automatic measurement technology, opened opportunity for continuous monitoring of fruit drying.[23] Comprehensive reviews of techniques and their applications have been provided.[24–26] Considerable efforts were made to study the quality changes of foodstuffs during drying using CVSs.[16,23] It was concluded that real-time CVS has a great potential in drying technology for material appearance monitoring.

This paper deals with the determination of proper conditions of HID–HAD for the production of dried kiwifruits. To the best of authors’ knowledge, little information is available in regards to improvement of kiwifruit hybrid drying, based on CVS monitoring of color and shrinkage. This knowledge is critically important for process enhancement. The objectives of this study were: (i) to investigate the effects of air velo-city, temperature, thickness of kiwifruit slices, and exposure/nonexposure to infrared (IR) radiation on color and shape changes and (ii) to determine the best conditions for kiwifruit drying, based on minimal drying time, shrinkage, and color changes.

Materials and methods

Materials

The kiwifruits (Actinidia deliciosa cv. Hayward) with initial moisture content of 82 � 1% wet basis were cold-stored at 5°C during seven-day period. For each experimental run, kiwifruits were peeled and then cut into slices perpendicular to the fruit axis by an on-fixture-mounted coping saw. Slices had an average diameter of 4.3 � 0.2 mm. In each experiment, one slice was used to determine the dimensionless refractive index immediately after removing a sample from refrigerator. In all experiments, refractive index of the kiwifruits was in the range of 13.95 � 0.25 indicating that fruit samples were uniformly ripened. To prevent enzymatic browning during the initial stage of drying, kiwifruit slices were pretreated by hot water blanching (75°C) for 5 min.[27]

Experimental setup

Figure 1 illustrates schematic of the lab-scale hybrid HID developed in the department of Biosystems Engin-eering at Ferdowsi University of Mashhad in Iran and

used for both HAD and HID drying processes. In fact, the dryer was designed such that convective or infrared drying could be performed individually or in combi-nation. The main parts of the hybrid dryer were: an electro-fan 2,000 m3/h, equipped with a 0.75 kW three-phase electromotor (Motogen, Iran), four 250 W IR lamps (Siccatherm, OSRAM, China), electric heating elements with total combined wattage of 7,000 W, an integrated control unit, a digital weighing scale (�0.01 g, A&D Co., Japan) connected to a PC for con-tinuously measuring sample’s weight, an air duct, a wire-meshed tray three integrated temperature and rela-tive humidity sensors (AM2303, Aosong Electronics Co., China). Two sensors were installed before and after the tray and the third one outside the dryer for measur-ing room temperature and humidity. The whole body of the dryer was thermally insulated with glass wool. The control unit was connected to a PC through RS-232 port. A microcontroller (ATmega16, Shenzhen Sunho-key Electronics Co., Ltd., China) was used to acquire air temperature, velocity, and relative humidity from sensors. The controller was programmed using MATLAB v7 (Mathworks, Inc, USA) to execute several tasks, such as reading measurements and control the fan and the heater elements. The airflow velocity was adjusted by changing the rotational speed of the fan through an inverter (model Hyundai N50, USA). The IR lamps were controlled with ON–OFF approach. The distance between the IR lamps and the tray with kiwifruit slices was 20 cm. The uniformity of infrared radiation on the tray was evaluated by a thermal camera (model NEC G120, Nippon Avionics, Japan). The amount of infrared radiation at the material surface was 0.3 W/cm2, which was comparable with the inten-sity reported by others.[28,29] The product temperature was measured by a K-type thermocouple.

The heater elements were controlled by pulse-width modulation (PWM) approach. In this method, air tem-perature is adjusted by subsequently altering the duty cycle of heater elements until the desirable airflow temperature is reached. The PWM technique allows controlling the heaters with various output levels and hence providing a smoother control of drying air temperature.[30] Before each drying run, the dryer was running for 20 min to achieve a stable steady-state condition.

Drying experiments

For this study, a total of 162 experimental combinations or treatments (three air temperatures � three air velocities � three sample thicknesses � two drying methods � three replicates) were conducted. The weight

710 M. H. NADIAN ET AL.

of samples was measured and stored every 30 s by a digital weighing scale connected to the PC through a RS-232 serial port and WinCT© (RsKey, Japan) software. The sample slices were dried up to a predeter-mined mass.

Calculation of moisture ratio

The moisture content (Mt) of kiwifruit slices was calculated during drying using Eq. (1):

Mt ¼Wt � Wd

Wd; ð1Þ

where Wt and Wd in grams (g) are the current weight and the dried weight of sample, respectively. The moist-ure ratio (MR) was calculated as:

MR ¼Mt � Me

Mo � Me; ð2Þ

where Me and Mo are the equilibrium moisture content and the initial moisture content, respectively. However,

the Me can be neglected for products having high initial moisture content. Therefore, Eq. (2) was simplified as[31]:

MR ¼Mt

Mo: ð3Þ

Computer vision system

Computer vision system consisted of a CCD color cam-era (Coolpix P510 model, Nikon Inc., Japan Nikon Co.) to continuously monitor surface area and color of slices and an illumination chamber composes of eight power LED lamps (5W, F-B4, Shenzhen Fuqiao Investment & Development Co., Ltd. China) with color temperature of 6,500 K placed 20 cm above the sample tray at an angle of 45° with respect to the sample plane to provide a uniform illumination (Fig. 2). To capture the high- quality images of the drying samples without effect from the ambient light, the camera was placed outside of the drying chamber, but lens was placed in the hole looking

Figure 1. Experimental dryer: (1) control units, (2) PC, (3) heating unit, (4) fan, (5) straightener, (6) balance, (7) illumination and imaging chamber (a cutaway of chamber is displayed), (8) inverter, (9) camera, (10) lighting lamp, (11) tray, (12) temperature sensors, (13) IR lamps.

DRYING TECHNOLOGY 711

directly toward drying tray (Fig. 2). A small 12-V fan was installed adjacent to the camera for cooling. The images of sliced fruit taken by a digital camera were sent through a Wi-Fi memory card (model Eyefi Mobi, Eyefi Co., USA) to the PC and then analyzed by the MATLAB software for online monitoring of the dried products. The surface shrinkage and color changes of slices were determined by digital analyzing the top view images of the slices on the tray captured every 30 s.

Color calibration The camera was calibrated manually through a trial and error method, to have proper color rendering. Color changes during drying were determined by extracting the RGB color information of kiwifruit segmented images, and then were converted to L*a*b* color space in two stages (RGB to XYZ and then XYZ to L*a*b*) based on the method comprehensively explained by Hosseinpour et al.[32] The reference white for calcu-lation of XYZ coordinates is according to CIE Standard D65. To calibrate the extracted L*a*b* color compo-nents, the L*, a*, and b* color values of 20 standard colored papers were measured by a colorimeter (Konica Minolta, model CR-410, Japan). These values were then used in a regression model against their corresponding values provided by the CVS in two conditions: with IR radiation and with no IR exposure, according to the method used by León et al.[33] These models are shown in Table 1. This procedure provided standard datasets for the comparison of color changes during different treatments in both HAD and HID modes of drying.

Image preprocessing

The processing of images was performed using an algorithm written in MATLAB. One of the first pre-processing steps was image segmentation. Our experi-mental settings allowed easy separation of object of interest, i.e., slices of kiwifruit, from the background. Kiwifruit slices had bright green color, while the back-ground (tray) was black, so segmentation was per-formed by thresholding the high-contrast image of green and blue difference (2G–2B) by Otsu’s method.[34]

This method chooses a threshold that minimizes the interclass variance of foreground and background classes. After binarizing images with Otsu’s method, morphological flood-fill operation was performed to fill the holes inside the regions of kiwi slices particularly due to the small black seeds. Figure 3 illustrates image processing steps followed for segmentation and feature extraction of some typical images at the beginning and end of HAD and HID drying.

Image feature extraction

Several features including the area and color of slices were extracted from the preprocessed images. The pro-jection area of slice (A) was calculated by the pixel count method. Area shrinkage was calculated as:

Area Shrinkage ð%Þ ¼A0 � A

A0; ð4Þ

where A0 and A are the initial and current projection areas of slice, respectively.

The calibrated L*, a*, and b* values of images were used to calculate the total amount of color changes during fruit drying as[32]:

DE ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

ðL� � L�0Þ2þ ða� � a�0Þ

2þ ðb� � b�0Þ

2q

; ð5Þ

where L�0, a�0, and b�0 represent the lightness, redness, and yellowness of a fresh sample, respectively.

Statistical analysis

The required drying time, total color change, and shrinkage of slices were recorded until the water content

Figure 2. A section view of illumination and imaging chamber: (1) camera, (2) power LED lighting lamp, and (3) IR lamps.

Table 1. Regression models for L*, a*, and b* calibration. IR lamp mode CIELAB Conversion function Correlation coefficient OFF L* Y ¼ 0.002X2 � 0.1605X þ 43.089 0.9851

a* Y ¼ � 0.0019X2 þ 1.3457X � 142.86 0.9663 b* Y ¼ 0.0003X2 þ 0.552X � 70.538 0.9453

ON L* Y ¼ 0.0029X2 � 0.4275X þ 56.143 0.9725 a* Y ¼ � 0.0063X2 þ 2.9142X � 304.25 0.9468 b* Y ¼ 0.0012X2 þ 0.3795X � 70.396 0.9298

IR, infrared.

712 M. H. NADIAN ET AL.

of the samples decreased to nearly MR ¼ 0.05. Statistical analysis of the results was carried using a 3 � 3 � 3 � 2 full factorial design (three temperatures � three air velocities � three thicknesses � two modes: infrared ON or OFF). Analysis of variance (ANOVA) was per-formed by the SAS software (version 9, SAS Institute Inc., Cary, NC, USA) to evaluate the significance of the effect of independent variables and their interactions on the required drying time, the total color change, and shrinkage of kiwifruit slices. Data were checked for nor-mality and homogeneity of residuals before ANOVA and were transformed by square root (for ΔE) and natu-ral logarithm (for drying time and area shrinkage) to satisfy the assumption of normality. Significant differ-ences among means were identified by least significant difference at the 0.05 level.

ALM clustering and improvement of kiwifruit drying

The main objective of this study was to select the best condition of kiwifruit drying (HID and/or HAD) among methods tested. To rank the treatments (drying settings), the drying time, the total color change, and the shrinkage of slices at MR ¼ 0.05 were considered as three separate categories and examined using cluster analysis. Clustering was performed with average linkage method (ALM) (SAS version 9, SAS Institute Inc., Cary, NC, USA) to identify similarities between a group members within each category.[35] Based on the range

of responses, a normalized value d (desirability) between 0 and 1 was allocated to each group member in such a way that the higher score represented more desirable response.

Then, a scoring method was chosen to determine the best treatment, which maximizes total score. For multiple responses, individual scores for each response were combined into total score, using total desirability function (D). Hence, the total score for each treatment was calculated using the following equation[36]:

D ¼Yn

i¼1di

!1=n

¼ ðdDryingTime � dDE � dAreaShrinkableÞ1=3: ð6Þ

The goal of improvement was to find a combination of drying factors that maximizes desirability function. In case, if more importance were given to color and shrinkage as compared to drying time, it would auto-matically eliminate HID drying from desirable settings. Therefore, initially all weights were considered of equal importance. Finally, the treatment with the highest desirability (D) was selected as the best.

Results and discussion

Effect of drying parameters on drying time

Figure 4 shows the effect of drying method and tem-peratures (A), slice thickness (B), and air velocity (C)

Figure 3. Consecutive steps for extracting foreground from background and finally image feature extraction: (a) original color image before drying, (b) grayscale image from “2G–2B,” (c) binary image, (d) binary image with filled holes, (e) final processed image at the beginning of drying, and (f) final processed image at the end of drying. Number 1, HAD drying method; number 2, HID drying method.

DRYING TECHNOLOGY 713

on the moisture kinetics. A general exponential trend for MR changes over time for all treatments has been observed. In general, at the beginning of drying process, when the material has still significant moisture content, the drying was limited with the rate of evaporation from the surface. At the second stage of drying, when moist-ure content fell below critical value, the major limitation was diffusion of the moisture from interior to the sur-face. Because diffusion may be hindered by product shrinkage,[1] case hardening,[21] or dried cell layers,[37]

the second stage was characterized with significantly smaller drying rate. Evaporation of the intercellular free

water took place in both stages of drying, but predomi-nantly at the second stage of drying. Effect of HID on drying rate at the final stages of drying could be associa-ted with the increase in IR reflectivity at low moisture contents.[10] Fig. 4a also clearly indicates that the infra-red was the key factor, affecting the drying kinetics at the end of drying. As shown in Table 2, the effects of all main independent variables and their interactions on drying time were significant. As also shown in Fig. 5, the water in all HID samples evaporated signifi-cantly faster than HAD samples. It is important to note that IR radiation has relatively small depth of pen-etration with direct energy transfer to the sample, thus creating a large vapor pressure differential between the center and surface. In all treatments, a higher drying temperature provides a faster drying process (Fig. 5). This trend was reported elsewhere, e.g., by Diamante et al.[38] and Orikasa et al.[22] for kiwifruits. Also, the drying time increased with a sample thickness (Figs. 4b and 5). It could be related to less amount of water, better mass and heat transfer, moisture and temperature gradi-ents, and larger exposure of surface area per volume in the thinner samples.[39]

The effect of air velocity on drying kinetics in both HAD and HID methods is shown in Figs. 4c and 5. It follows that HAD drying time often significantly decreased with the increase in air velocity. This was expected as Kaya et al.[40] stated that air velocity accel-erates moisture diffusion and heat transfer of kiwifruit samples. However, the opposite effect was observed in HID drying with the highest drying rate recorded at the lowest air velocity of 0.5 m/s. The explanation of this fact was provided by Nowak and Lewicki.[10] Twice faster IR drying by reducing air velocity from 1.5 to 0.5 m/s, they explained by the increase in heat transfer. In fact, in the absence of airflow, the temperature of product surface exposed to infrared would continuously increase during drying. As long as the temperature of product is higher than the air temperature, the heat is transferred from the surface to the air. In other words, the airflow removes some of the heat supplied by IR radiation to the surface and thus cooling the surface. This cooling effect increases with air velocity and nega-tively affects drying rate. Another hypothesis of drying rate decrease is due to the changes of vapor pressure.[41]

Effect of drying parameters on shrinkage

The area shrinkage of kiwifruits versus their MRs during different drying treatments is shown in Fig. 6a–c. It is well known that shrinkage develops in samples with the water removal. When water is removed from the slices, an unbalance pressure leads

Figure 4. Kinetics of MR with HID and HAD drying methods for kiwifruits at different conditions: (a) different temperatures and constant air velocity of 1 m/s and constant thickness of 5 mm, (b) different thicknesses and constant air velocity of 1.5 m/s and constant temperature of 70°C, and (c) different air velocities, and constant thickness of 7 mm and constant temperature of 60°C.

714 M. H. NADIAN ET AL.

to material shrinkage, shape change, and occasional cracking. As could be seen from Fig. 6, during the initial phase of drying, when the drying rate is high, the shrinkage was also high. At high moisture contents, the slices were in the rubbery state with the high mobility of solid matrix.[42] In contrast, at low moisture contents the material changed from rubbery to glassy state, which remarkably decreased the shrinkage.[43]

Table 2 shows significant effects of all main independent variables and their interactions on area shrinkage. The effect of air temperature on slice area shrinkage at constant air velocity of 1.0 m/s and

thickness of 5 mm is presented in Fig. 6a. In air convec-tive drying (HAD), shrinkage often increased with tem-perature, which corresponds to earlier research findings for juicy fruits drying.[44] In fruits, glass-transition tem-perature is determined by concentration of sugars (glu-cose, fructose, and sucrose) and biopolymers[45] and moisture content. With the decrease in moisture con-tent, the product temperature (Tp) increases and comes closer to the glass-transition temperature (Tg), thus decreasing the shrinkage. It can be said that the greater the difference (Tp � Tg), the higher shrinkage can happen.[42]

Table 2. ANOVA results for the effect of temperature, air velocity, thickness of sample, and infrared ray on drying time, area shrinkage, and DE during drying up to moisture ratio of 0.05.

Source of variance DF

Drying time Area shrinkage DE

SS Pr>F SS Pr>F SS Pr>F

T 2 2.16 <0.0001 0.06 <0.0001 0.61 0.0004 V 2 0.49 <0.0001 0.03 <0.0001 0.46 <0.0001 T � V 4 0.44 <0.0001 0.12 <0.0001 0.92 <0.0001 T 2 30.59 <0.0001 11.96 <0.0001 8.12 <0.0001 T � t 4 0.12 <0.0001 0.20 <0.0001 2.38 <0.0001 V � t 4 0.20 <0.0001 0.58 <0.0001 1.66 <0.0001 T � V � t 8 0.17 <0.0001 0.21 <0.0001 1.28 <0.0001 IR 1 84.59 <0.0001 0.06 <0.0001 21.00 <0.0001 T � IR 2 0.60 <0.0001 0.45 0.0006 3.76 <0.0001 V � IR 2 4.08 <0.0001 0.06 <0.0001 1.16 <0.0001 T � V � IR 4 0.05 <0.0001 0.27 <0.0001 0.66 <0.0001 t � IR 2 0.04 <0.0001 0.27 <0.0001 4.15 <0.0001 T � t � IR 4 0.36 <0.0001 0.05 <0.0001 3.07 <0.0001 V � t � IR 4 0.09 <0.0001 0.51 <0.0001 0.58 <0.0001 T � V � t � IR 8 0.14 <0.0001 0.29 <0.0001 0.60 <0.0001 Error 108 0.11 0.10 0.63 CV 0.78 1.03 2.99

ANOVA, analysis of variance; DF, degree of freedom; SS, sum of square; T, temperature; V, air velocity; t, thickness; IR, infrared ray. Bold numbers indicate significant effect.

Figure 5. The effect of drying method, temperature, air velocity, and thickness of slice on drying time, for drying up to moisture ratio of 0.05 (LSD ¼ 5.43; P > 0.05).

DRYING TECHNOLOGY 715

In HID drying, the air temperature had no meaning-ful effect on shrinkage (Fig. 7). It might be attributed to major effect of radiative heat transfer, which was pre-dominant over the effect of air temperature. When material is exposed to high temperature or IR radiation, rapid drying may lead to surface crust formation. Extremely low moisture content of outer shell induces early glass transition that stops the shrinkage of the still rubbery inner part.[43] The combination of impermeable crust with extreme surface temperature may result in rapid moisture vaporization and significant internal pressure, causing product expansion/puffing.[46] In this case, mechanical stress may provoke surface cracking. If

cracked, the crust may lose impermeability and the degree of product expansion may decrease.[46]

The effect of drying method on shrinkage in our study was somehow complicated. In most treatments, more shrinkage was observed with HID method com-pared with HAD (Figs. 6b and 7). It could be explained by superimposing effect of air-convective and radiative heating, which resulted in large difference between pro-duct and glass-transition temperatures. We could assume that mobility of the matrix was not enough to support the structure of the solid material and collapse took place.[47] On the other hand, if drying conditions do not allow a phase transition in the outer layer, shell formation is not observed even at high drying rates.[43]

The results also indicated that the thicker the slices, the more shrinkage occurs (Figs. 6b and 7). As pre-viously mentioned, longer drying gives more time for the product to shrink.[48] Also, in thicker slices, the outer surface layer dries faster than the inner parts and therefore, volumetric shrinkage is very uneven. Thus, both drying rate and glass transition have pro-found effects on the progression of the shrinkage. The above factors could be the reasons of the complexity of shrinkage behavior of the samples with different thickness.

The effect of air velocity on shrinkage was not consistent (Figs. 6c and 7). The convective flow of air removes the moisture from the surface and cooling the surface, which results in mass transfer.[49] Hence, surface moisture depends on drying time and heating mode, wide ranges of Tg and Tp may occur. As pre-viously mentioned, this can be reason for different shrinkage at different air velocity, which makes it difficult to interpret.

Effect of drying parameters on total color changes

Color changes were calculated from L*a*b* measure-ments with Eq. (5). In our study, effects of all main independent variables and their interactions on color changes ΔE were significant (Table 2). Effects of various drying treatments on the color changes in kiwifruit slices as a function of MR are presented in Fig. 8a–c. A general trend of increasing ΔE with removing moist-ure from samples was observed. It is important to note that color changes are more pronounced with HID as compared to HAD method. Maskan[20] reported that HAD of kiwifruits had little ΔE compared to microwave and hybrid hot air–microwave methods. Therefore, HAD is an appropriate method for preserving the color of this fruit. However, the maximum color difference (ΔE) was obtained for slices dried by HID method

Figure 6. Relationships between area shrinkage and MR of kiwifruit slices for HID and HAD drying methods at different con-ditions: (a) different temperatures and constant air velocity of 1.0 m/s and thickness of 5 mm, (b) different thicknesses and constant air velocity of 1.5 m/s and temperature of 70°C, and (c) different air velocities and constant thickness of 7 mm and temperature of 60°C.

716 M. H. NADIAN ET AL.

and at 70°C air temperature, probably due to the none-nzymatic browning (Maillard reaction). As the Maillard reaction accelerated, the interaction between sugars and amino acids at higher temperature leads to formation of

brown compounds. In general, the absorption of IR heat by the surface of the material under drying is the main reason for the higher ΔE in HID-dried samples. None-nzymatic browning was the most intensive at the last

Figure 8. Relationships between ΔE and MR of kiwifruit slices for both HID and HAD drying methods at different conditions: (a) various temperatures, air velocity of 1 m/s, and thickness of 5 mm; (b) various thicknesses, air velocity of 1.5 m/s, and temperature of 70°C; and (c) and various air velocities, thickness of 7 mm, and temperature of 60°C.

Figure 7. The effect of drying method, temperature, air velocity, and thickness of slice on the area shrinkage, for drying up to moisture ratio of 0.05 (LSD ¼ 0.99; P > 0.05).

DRYING TECHNOLOGY 717

stage of drying, when the temperature of the surface was the highest and the moisture content reached minimum. In this regard, Garza et al.[50] indicated a direct relation-ship between temperature and nonenzymatic browning rate.

Interestingly, with some treatments of HAD and HID dryings, the ΔE of samples at 50°C temperature was relatively higher than at 60 or 70°C. In this regard, Orikasa et al.[51] reported that the maximum ΔE was obtained for kiwifruit slices dried by HAD at 50°C air temperature. This fact could be explained by longer dry-ing time. The same phenomenon was observed by Nadian et al.[16] and Baini and Langrish[52] for drying of apple and banana slices, respectively. In other words, in our experiments the effect of drying time on none-nzymatic browning was more significant than the effect of air temperature. For example, in HAD method the slices with thickness of 3 mm and temperature of 70°C had less color change, which could be attributed to shorter drying time. In this regard, to get the highest quality product, a compromise between the drying temperature and drying time should be found.

In most cases, according to Figs. 8b and 9, increasing the thickness of kiwifruit slices results in noteworthy darker color. In this regard, Orikasa et al.[51] also observed a high ΔE (greater than 12) when kiwifruit samples of 10-mm thickness, dried by HAD at 50–70°C temperatures. The decrease in air velocity in HAD and increase in HID drying method increased

color deterioration (Fig. 8c). These results also could be related to duration of drying and Maillard reaction, which govern the darkening of the final product. Therefore, in HID method, care should be taken for applying adequate air velocity and IR exposure to get final product with the least color change. Also, the thinner slices (3 mm) fairly prevent the quality deterio-ration in term of color.

Cluster analysis

Figure 10 presents the results of ALM cluster analysis for 54 drying treatments presented in Table 3 with respect to their impact on drying time (Fig. 10a), area shrinkage (Fig. 10b), and color changes (Fig. 10c). These tree diagrams indicate the cluster membership of each treatment in the cluster tree. The name of cluster and average distance between clusters is displayed on the horizontal and vertical axes, respectively. Also, each of the binary trees has leaves, branches, and roots. The leaves are represented by row numbers in Table 3. A horizontal dashed line is drawn on any of the trees so that the best clustering separation obtained among all 54 treatments. Therefore, branches separated by the dashed line have the same characteristics. For example, in Fig 10a, leaves 1, 2, 5, and 9, i.e., HID (70°C–1.5 m/s–7 mm), HID (70°C–1.0 m/s–7 mm), HID (60°C–1.0 m/s–7 mm), and HID (50°C–0.5 m/s– 7 mm) are located on the same branch of tree and

Figure 9. The effect of drying method, temperature, air velocity, and thickness of slice on the ΔE, for drying up to moisture ratio of 0.05 (LSD ¼ 0.72; P > 0.05).

718 M. H. NADIAN ET AL.

therefore, there were no differences among them in regards of drying time. Finally, each branch or leaf sepa-rated by the dashed line, received a score between 0 and

1 based on its average value. Then, all partial scores (d) obtained for each treatment were calculated by Eq. (6) and presented as a total desirability score (D) in

Figure 10. Tree diagram of (a) drying time, (b) area shrinkage, and (c) ΔE clusters versus average distance between clusters.

DRYING TECHNOLOGY 719

Table 3. According to this table, in HAD drying, the combination of the highest temperature with maximum air velocity is the most desirable. These results were consistent with those of Sturm et al.[18] for HAD drying of apples. On the other hand, in HID drying the highest temperature with minimum air velocity showed reason-ably good result. In all cases, HID drying required less drying time in comparison to HAD. Also, the thinner samples could well prevent the quality deterioration because of shorter drying time. These results indicate that the two of HAD treatments including the air temperature of 70°C, velocity of 1.5 m/s, and the thickness of 3 mm (row number: 46); and the air temperature of 50°C, velo-city of 1.0 m/s, and thickness of 3 mm (row number: 53) had the highest desirability of final quality among all 54 treatments for kiwifruit drying. Although the treatment with row number 53 ranked as the highest desirability in terms of appearance quality by CVS (the desirability of shrinkage and color changes was equal to 1, it needed more drying time in comparison to row number 46 (Table 3). It was reported that drying time might nega-tively affect the ratio of antioxidant activity, the residual ratio of AsA and the hardening ratio of kiwifruit.[51]

It could be concluded that air temperature 70°C, thickness 3 mm, air velocity 1.5 m/s among all HAD treatments, and the same air temperature 70°C, thickness 3 mm, air velocity 0.5 m/s among all HID treatments had the highest desirability and therefore, were selected as the best conditions for kiwifruit drying.

Hybrid kiwifruit drying

Considering the results and discussion provided in the previous section, two drying treatments were chosen as the best conditions for kiwifruit drying among all treatments. Although HID method could well minimize the drying time, it did not fulfill the expectations for the “end-product quality”. Indeed, as it could be seen in Figs. 6 and 8, most of color and shrinkage deterioration of 3-mm slices for the HID method occurs almost after MR ¼ 0.7 (almost 5 min after the drying starts) when the Tp increased (Fig. 11d). Hence, to minimize ΔE and shrinkage, a significantly lower temperature needs to be chosen at the end of drying process. Consequently, to have the best quality of the product with the shortest drying time, we used a hybrid approach which is a combination of the best conditions of HID and HAD methods. In this way, the dryer operated in HID drying mode at air temperature of 70°C and air velocity of 0.5 m/s for the first 5 min of drying and then switched to HAD mode at the same air temperature 70°C with air velocity of 1.5 m/s to continue drying. The results of this hybrid approach compared with either HAD or HID are shown in Fig. 11. As seen in this figure, the hybrid approach could minimize the drying time in comparison with the HAD method (Fig. 11a) with a lit-tle change in color (Fig. 11c). In other words, this approach accelerated drying process, while prevented significant deterioration of product quality in regards

Table 3. The desirability (d) of different drying methods based on tree diagrams of drying time, area shrinkage, and ΔE clusters. Number 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Drying method HID HID HID HID HID HID HID HID HID HID HID HID HID HID HID HID HID HID

Temperature (°C) 70 70 70 60 60 60 50 50 50 70 70 70 60 60 60 50 50 50 Air velocity (m/s) 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 Thickness (mm) 7 7 7 7 7 7 7 7 7 5 5 5 5 5 5 5 5 5 dDryingTime 0.80 0.80 0.85 0.75 0.80 0.85 0.70 0.75 0.80 0.90 0.90 0.95 0.90 0.90 0.95 0.90 0.90 0.95 dAreaShrinkable 0.20 0.20 0.25 0.25 0.10 0.20 0.20 0.15 0.00 0.55 0.75 0.65 0.70 0.75 0.60 0.70 0.85 0.35 dΔE 0.00 0.13 0.13 0.06 0.19 0.44 0.38 0.31 0.25 0.13 0.06 0.25 0.69 0.63 0.56 0.69 0.44 0.38 Total desirability (D) 0.00 0.28 0.30 0.22 0.25 0.42 0.38 0.33 0.00 0.40 0.34 0.54 0.76 0.75 0.68 0.76 0.70 0.50 Number 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 Drying method HID HID HID HID HID HID HID HID HID HAD HAD HAD HAD HAD HAD HAD HAD HAD Temperature (°C) 70 70 70 60 60 60 50 50 50 70 70 70 60 60 60 50 50 50 Air velocity (m/s) 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 Thickness (mm) 3 3 3 3 3 3 3 3 3 7 7 7 7 7 7 7 7 7 dDryingTime 1.00 1.00 1.00 1.00 1.00 1.00 0.95 0.95 1.00 0.35 0.25 0.10 0.30 0.20 0.05 0.15 0.05 0.00 dAreaShrinkable 0.80 0.75 0.90 0.60 0.75 0.90 0.60 0.75 1.00 0.35 0.10 0.05 0.45 0.40 0.30 0.15 0.10 0.30 dΔE 0.38 0.31 0.50 0.44 0.31 0.31 0.38 0.44 0.19 0.69 0.69 0.69 0.69 0.69 0.63 0.56 0.31 0.50 Total desirability (D) 0.67 0.61 0.77 0.64 0.61 0.65 0.60 0.68 0.57 0.44 0.26 0.15 0.45 0.38 0.21 0.23 0.12 0.00 Number 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 Drying method HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD HAD Temperature (°C) 70 70 70 60 60 60 50 50 50 70 70 70 60 60 60 50 50 50 Air velocity (m/s) 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 1.5 1.0 0.5 Thickness (mm) 5 5 5 5 5 5 5 5 5 3 3 3 3 3 3 3 3 3 dDryingTime 0.70 0.60 0.40 0.65 0.45 0.30 0.40 0.35 0.20 0.90 0.70 0.60 0.75 0.70 0.55 0.65 0.60 0.50 dAreaShrinkable 0.55 0.60 0.50 0.65 0.65 0.60 0.50 0.70 0.75 0.95 0.85 0.75 0.85 0.85 0.90 0.90 1.00 0.95 dΔE 0.81 0.56 0.50 0.69 0.50 0.38 0.56 0.69 0.31 1.00 1.00 0.94 0.88 0.75 0.63 1.00 1.00 0.88 Total desirability (D) 0.68 0.59 0.46 0.66 0.53 0.41 0.48 0.55 0.36 0.95 0.84 0.75 0.82 0.76 0.68 0.84 0.85 0.75

HID, hot air-infrared drying; HAD, hot air drying. Values in bold type are associated with the best conditions for kiwifruit drying.

720 M. H. NADIAN ET AL.

to color and shrinkage. Maskan[21] reported that hybrid hot air–microwave drying, in addition to reducing dry-ing time, resulted in substantial improvement in quality attributes, such as shrinkage. As seen in Fig. 11d, this method could well control the temperature of the slices, which is an important factor during drying.[14,15]

Also, increasing the product temperature at the first stage of drying is a desirable condition to accelerate dry-ing process. Based on the results obtained, a stepwise process could be used for the control of the drying process.

Conclusion

The CVS was successfully applied for real-time monitoring of the shrinkage and color changes of kiwi-fruit slices in a hybrid HAD–HID drying. The effect of reddish reflection of infrared lamps on the image color was corrected using experimentally established L*a*b* calibration procedure. The results showed that slices of small thickness (<5 mm) resulted in shorter drying time and better quality. HAD drying at air temperature of 70°C and air velocity of 1.5 m/s could prevent quality degradation compared to other treatments. Although HID drying considerably decreased drying time, the deterioration of product quality in terms of color and shrinkage was significant. It was found that increasing the product temperature accelerates drying process

without damaging quality at the first stage of drying. Therefore, the two-stage hybrid method with HID at the first stage and HAD at the second stage of drying was proposed. In general, this method reduces the drying time, while maintaining the quality of dried products at a reasonable level. In addition, this method provided the benefit of temperature control of the sam-ple surface, which is an important factor during drying. Desirability function allowed us to find conditions of hybrid HID–HAD drying for simultaneous minimiza-tion of drying time, color changes and the shrinkage. However, further studies are necessary to develop a real-time monitoring and feedback control of infrared lamps, temperature, and air velocity, so that the advantages of both HAD and HID would be incorpor-ated in the drying process to acquire the best-quality products.

Funding

Authors would like to extend their appreciations for the financial support provided by the Ferdowsi University of Mashhad.

References

[1] Golestani, R.; Raisi, A.; Aroujalian, A. Mathematical modeling on air drying of apples considering shrinkage

Figure 11. Trends of MR (a), area shrinkage (b), color change ΔE (c), and product temperature (d), with HID (70°C–0.5 m/s–3 mm), HAD (70°C–1.5 m/s–3 mm), and hybrid drying approach for kiwifruits.

DRYING TECHNOLOGY 721

and variable diffusion coefficient. Drying Technology 2013, 31(1), 40–51.

[2] Wang, J. A single-layer model for far-infrared radiation drying of onion slices. Drying Technology 2002, 20(10), 1941–1953.

[3] Wray, D.; Ramaswamy, H.S. Novel concepts in micro-wave drying of foods. Drying Technology 2015, 33(7), 769–783.

[4] Li, R.; Huang, L.; Zhang, M.; Mujumdar, A.S.; Wang, Y.C. Freeze drying of apple slices with and without application of microwaves. Drying Technology 2014, 32(15), 1769–1776.

[5] Garcia-Perez, J.V.; Carcel, J.A.; Riera, E.; Rosselló, C.; Mulet, A. Intensification of low-temperature drying by using ultrasound. Drying Technology 2012, 30(11–12), 1199–1208.

[6] Ciurzyńska, A.; Piotrowski, D.; Lenart, A.; Łukasik, P. Sorption properties of vacuum-dried strawberries. Drying Technology 2012, 30(8), 850–858.

[7] Kowalski, S.J.; Mierzwa, D. Hybrid drying of red bell pepper: Energy and quality issues. Drying Technology 2011, 29(10), 1195–1203.

[8] Kudra, T. Energy aspects in drying. Drying Technology 2004, 22(5), 917–932.

[9] Riadh, M.H.; Ahmad, S.A.B.; Marhaban, M.H.; Soh, A.C. Infrared heating in food drying: An overview. Drying Technology 2015, 33(3), 322–335.

[10] Nowak, D.; Lewicki, P.P. Infrared drying of apple slices. Innovative Food Science & Emerging Technologies 2004, 5(3), 353–360.

[11] Mujumdar, A.S.; Devahastin, S. Mujumdar’s Practical Guide to Industrial Drying; Exergex Corporation: Canada, 2000.

[12] Chou, S.; Chua, K. New hybrid drying technologies for heat sensitive foodstuffs. Trends in Food Science & Technology 2001, 12(10), 359–369.

[13] Feng, L.; Zhang, M.; Adhikari, B. Effect of water on the quality of dehydrated products: A review of novel char-acterization methods and hybrid drying technologies. Drying Technology 2014, 32(15), 1872–1884.

[14] Sturm, B.; Vega, A.-M.N.; Hofacker, W.C. Influence of process control strategies on drying kinetics, colour and shrinkage of air dried apples. Applied Thermal Engineering 2014, 62(2), 455–460.

[15] Vega, A.-M.N.; Sturm, B.; Hofacker, W. Simulation of the convective drying process with automatic control of surface temperature. Journal of Food Engineering 2016, 170, 16–23.

[16] Nadian, M.H.; Rafiee, S.; Aghbashlo, M.; Hosseinpour, S.; Mohtasebi, S.S. Continuous real-time monitoring and neural network modeling of apple slices color changes during hot air drying. Food and Bioproducts Processing 2015, 94, 263–274.

[17] Aghbashlo, M.; Hosseinpour, S.; Ghasemi- Varnamkhasti, M. Computer vision technology for real-time food quality assurance during drying process. Trends in Food Science & Technology 2014, 39(1), 76–84.

[18] Sturm, B.; Hofacker, W.C.; Hensel, O. Optimizing the drying parameters for hot-air-dried apples. Drying Technology 2012, 30(14), 1570–1582.

[19] Mujumdar, A.S. Drying Technology in Agriculture and Food Sciences; Science Publishers, Inc.: Plymouth, UK, 2000.

[20] Maskan, M. Kinetics of colour change of kiwifruits during hot air and microwave drying. Journal of Food Engineering 2001, 48(2), 169–175.

[21] Maskan, M. Drying, shrinkage and rehydration characteristics of kiwifruits during hot air and micro-wave drying. Journal of Food Engineering 2001, 48(2), 177–182.

[22] Orikasa, T.; Wu, L.; Shiina, T.; Tagawa, A. Drying characteristics of kiwifruit during hot air drying. Journal of Food Engineering 2008, 85(2), 303–308.

[23] Martynenko, A.I. Computer-vision system for control of drying processes. Drying Technology 2006, 24(7), 879–888.

[24] Wu, D.; Sun, D.-W. Colour measurements by computer vision for food quality control – A review. Trends in Food Science & Technology 2013, 29(1), 5–20.

[25] Brosnan, T.; Sun, D.-W. Improving quality inspection of food products by computer vision – A review. Journal of Food Engineering 2004, 61(1), 3–16.

[26] Davies, E. The application of machine vision to food and agriculture: A review. The Imaging Science Journal 2009, 57(4), 197–217.

[27] Nadian, M.H.; Abbaspour-Fard, M.H.; Sadrnia, H.; Golzarian, M.R.; Tabasizadeh, M. Optimal pretreatment determination of kiwifruit drying via online monitoring. Journal of the Science of Food and Agriculture 2016, 96(14), 4785–4796.

[28] Supmoon, N.; Noomhorm, A. Influence of combined hot air impingement and infrared drying on drying kinetics and physical properties of potato chips. Drying Technology 2013, 31(1), 24–31.

[29] Motevali, A.; Minaei, S.; Khoshtaghaza, M.H.; Amirnejat, H. Comparison of energy consumption and specific energy requirements of different methods for drying mushroom slices. Energy 2011, 36(11), 6433–6441.

[30] Javanmard, M.; Abbas, K.; Arvin, F. A microcontroller- based monitoring system for batch tea dryer. Journal of Agricultural Science 2009, 1(2), 101–106.

[31] Doymaz, İ. An experimental study on drying of green apples. Drying Technology 2009, 27(3), 478–485.

[32] Hosseinpour, S.; Rafiee, S.; Mohtasebi, S.S.; Aghbashlo, M. Application of computer vision technique for on-line monitoring of shrimp color changes during drying. Journal of Food Engineering 2013, 115(1), 99–114.

[33] León, K.; Mery, D.; Pedreschi, F.; León, J. Color measurement in L∗a∗b∗ units from RGB digital images. Food Research International 2006, 39(10), 1084–1091.

[34] Otsu, N. A threshold selection method from gray-level histograms. IEEE Transactions on Systems, Man, and Cybernetics 1979, 9(1), 62–66.

[35] Seifoddini, H.K. Single linkage versus average linkage clustering in machine cells formation applications. Computers & Industrial Engineering 1989, 16(3), 419–426.

[36] Costa, N.R.; Lourenço, J.; Pereira, Z.L. Desirability func-tion approach: A review and performance evaluation in adverse conditions. Chemometrics and Intelligent Laboratory Systems 2011, 107(2), 234–244.

722 M. H. NADIAN ET AL.

[37] Dıaz, G.R.; Martınez-Monzo, J.; Fito, P.; Chiralt, A. Modelling of dehydration-rehydration of orange slices in combined microwave/air drying. Innovative Food Science & Emerging Technologies 2003, 4(2), 203–209.

[38] Diamante, L.; Durand, M.; Savage, G.; Vanhanen, L. Effect of temperature on the drying characteristics, colour and ascorbic acid content of green and gold kiwi-fruits. International Food Research Journal 2010, 17, 441–451.

[39] Sacilik, K.; Elicin, A.K. The thin layer drying character-istics of organic apple slices. Journal of Food Engineering 2006, 73(3), 281–289.

[40] Kaya, A.; Aydın, O.; Kolaylı, S. Effect of different drying conditions on the vitamin C (ascorbic acid) content of Hayward kiwifruits (Actinidia deliciosa Planch). Food and Bioproducts Processing 2010, 88(2–3), 165–173.

[41] Defraeye, T.; Nicolaï, B.; Mannes, D.; Aregawi, W.; Verboven, P.; Derome, D. Probing inside fruit slices during convective drying by quantitative neutron imaging. Journal of Food Engineering 2016, 178, 198–202.

[42] Katekawa, M.E.; Silva, M.A. On the influence of glass transition on shrinkage in convective drying of fruits: A case study of banana drying. Drying Technology 2007, 25(10), 1659–1666.

[43] Mayor, L.; Sereno, A.M. Modelling shrinkage during convective drying of food materials: A review. Journal of Food Engineering 2004, 61(3), 373–386.

[44] Kurozawa, L.E.; Hubinger, M.D.; Park, K.J. Glass transition phenomenon on shrinkage of papaya during convective drying. Journal of Food Engineering 2012, 108(1), 43–50.

[45] Vega-Gálvez, A.; Ah-Hen, K.; Chacana, M.; Vergara, J.; Martínez-Monzó, J.; García-Segovia, P.; Lemus- Mondaca, R.; Di Scala, K. Effect of temperature and air velocity on drying kinetics, antioxidant capacity, total phenolic content, colour, texture and microstructure of apple (var. Granny Smith) slices. Food Chemistry 2012, 132(1), 51–59.

[46] Rahman, M.S. Toward prediction of porosity in foods during drying: A brief review. Drying Technology 2001, 19(1), 1–13.

[47] Champion, D.; Le Meste, M.; Simatos, D. Towards an improved understanding of glass transition and relaxa-tions in foods: Molecular mobility in the glass transition range. Trends in Food Science & Technology 2000, 11(2), 41–55.

[48] Ratti, C. Shrinkage during drying of foodstuffs. Journal of Food Engineering 1994, 23(1), 91–105.

[49] Hebbar, H.U.; Vishwanathan, K.; Ramesh, M. Development of combined infrared and hot air dryer for vegetables. Journal of Food Engineering 2004, 65(4), 557–563.

[50] Garza, S.; Ibarz, A.; Pagán, J.; Giner, J. Non-enzymatic browning in peach puree during heating. Food Research International 1999, 32(5), 335–343.

[51] Orikasa, T.; Koide, S.; Okamoto, S.; Imaizumi, T.; Muramatsu, Y.; Takeda, J.-I.; Shiina, T.; Tagawa, A. Impacts of hot air and vacuum drying on the quality attributes of kiwifruit slices. Journal of Food Engineering 2014, 125, 51–58.

[52] Baini, R.; Langrish, T.A.G. Assessment of colour development in dried bananas – Measurements and implications for modelling. Journal of Food Engineering 2009, 93(2), 177–182.

DRYING TECHNOLOGY 723