Indirect determination of moisture using biospeckle technique¹

Determinação indireta do teor de umidade utilizando a técnica de biospeckle

Abstract

The classification and selection of agricultural products are very important to assure post-harvest quality processes. However, most of physical and chemical analyses are destructive. It means those products turn useless after these analyses. Moisture is one of the most important parameters to assure the storage quality and as well to facilitate product processing. This research offers an alternative method to determine that parameter through an indirect and non-destructive way, using the image analysis based on the Biospeckle phenomenon. This method consists in get a 10 seconds video of a laser focused on the object. This video is divided in frames, using a computational routine in Matlab and, then, the sequence of images is processed using the computational program named ImageJ. The results are expressed in terms of modified correlation matrix and Moment of Inertia (MI). At the same time, a moisture analysis is carried out using the traditional methodology employing conventional heater equipment set at 130 ºC. By correlating MI and moisture values, it was possible to determine the moisture content by means of an indirect method. To this research, it was employed grains of Moringa oleifera.

Key words:
Biospeckle; Laser; Moisture; Moringa oleifera

Resumo

A classificação e a seleção de produtos agrícolas são de grande importância para garantir a qualidade dos processos de pós-colheita. No entanto, a maioria das análises físicas e químicas são destrutivas e isso significa que os produtos são inutilizados após essas análises. Sendo a umidade um dos mais importantes parâmetros para garantir a qualidade de armazenamento, bem como facilitar o processamento do produto, esse trabalho propõe um método alternativo para determinar esse parâmetro através de forma indireta e não-destrutiva, utilizando a análise de imagens a partir do fenômeno Biospeckle. Esse método consiste em obter um vídeo de 10s no qual o objeto é incidido por um laser. Esse vídeo é dividido em frames, utilizando rotina computacional no Matlab e, então, a sequência de imagens é processada, utilizando-se o programa computacional, ImageJ. Os resultados foram expressos em termos da matriz de correlação modificada e Momento de Inércia (MI). Em paralelo à análise de imagens, foi feita a análise de umidade através do método tradicional, empregando uma estufa a 130 ºC. Correlacionando MI e os valores de umidade, foi possível determinar a umidade pelo método indireto. Para esse trabalho foram empregados grãos de Moringa oleifera.

Palavras-chave:
Biospeckle; Laser; Umidade; Moringa oleifera

Introduction

In order to guarantee fruits and vegetables quality to market, it’s necessary to develop more accurate selection and sorting techniques. For this reason, an important research line explores nondestructive techniques.

The pertinent literature discloses several methods associated to quality determination that includes moisture content determination, seed viability, X-rays methods among others. Biospeckle techniques, which give tissue vitality level, are based on interference phenomenon that carries the information of the material. Guedes et al. (2014) evaluated kefir grains viability during the beverage production. Other authors aimed assess the shelf time of other products: Amaral et al. (2013) observed in beef color and the WBSF parameter during 21 days, Arefi et al. (2013) used biospeckle to compare fresh, semi-mealy and mealy apple and Alves, Braga and Vilas Boas (2013) identified respiration rate and water activity change in fresh-cut carrots.

This vitality is closely associated to tissue water content and is related to metabolic process intensity, which, in turn, influences the temporal variation of the interference Biospeckle (RODRIGUES et al., 2005).

The optical interference phenomena can be noted when non-biological phenomenon or a diffusing object are lighting with the laser, and it’s called dynamic speckle (RABELO, 2000).

In this research, the Biospeckle technique was evaluated using the Moment of Inertia (MI). It is based on speckle temporal variation. To quantify this variation, it was used the STS (Time History Speckle Pattern).

To distinguish the different levels of activity from the STS images, it becomes necessary to transform this image into a number or a quantification index (RABELO, 2000). Some calculations are necessary to this. Those related the probability of pixel intensity occurrence in each position of STS, resulting in a modified occurrence matrix (MOM).

According to Rabelo (2000), the occurrence values show how many times the intensity i is followed by the intensity j at the STS in the time sequence. If this intensity didn’t change over time, just the main diagonal will have non-null values of occurrence matrix, what characterize the material as a low activity material. On the other hand, we have a high activity material when the intensity changes along the time and, then, some non-null values begin to appear out of the main diagonal.

It can be applied the concept of Moment of Inertia of second order to reach a numeric value to quantify the differences between two distinct images. This calculation is applied on the MOM concerning its main diagonal in direction of lines, constituting the intensity dispersion module (MDI), also called in literature as MI (RABELO, 2000).

Based on that information, the present research was aimed to develop an alternative and indirect method to classify biological products based on its moisture, without destruct it.

It was used some grains of Moringa oleifera to develop that alternative method. It was used grains and not seeds, because in that case the objective was analyze just the influence of the water in the material.

The selection of Moringa oleifera is closed associated to its importance to many fields of knowledge. Particularly in water treatment, those grains can be used as an alternative to replace or to work together with the synthetic aluminum coagulant. But it’s important to have low level of moisture.

Material and methods

It was used Moringa oleifera grains for this research work. The grains were dried in the oven for 2 hours in 130 ºC to ensure that all the grains were kept dry (the initial average moisture was 7%), as it was described at MAPA (BRASIL, 2009).

After this, the process of grain humidification was carried, using a 0.5% of salt solution (NaCl). Some shelled grains were allocated in a container and then it was added the salt solution, as is showing in Figure 1.

Figure 1
The Moringa oleifera grains before (a) and after (b) the humidification

The grains were weighed to calculate the moisture in some time intervals as: 0h, 0.25 h, 0.5 h, 0.75 h, 1 h, 1.5 h, 2 h, 2.5 h, 3 h, 4 h, 5 h, 6 h, 7 h, 21 h, 23 h and 27 h. At these times, it was made the 10 seconds videos to the Biospeckle image.

The moistures were determinate by using the Equation 1, where: m_ai is the weight of water in the initial time; m₀ is the inicial weight (dry grains + m_ai); and m_a is the weight of water absorved by the grain.

(1) $\%\mathit{U}=\raisebox{1ex}{${\mathit{m}}_{\mathit{a}}$}\!\left/ \!\raisebox{-1ex}{$\left({\mathit{m}}_{\mathit{o}}-{\mathit{m}}_{\mathit{ai}}\right)·100\%$}\right.$

To analysis of Biospeckle, it was used the method named “Moment of Inertia” (MI). The sample was indirectly lighted with a red diode laser (632 nm), which had 10 mW of power. The experimental setup is showing in Figure 2.

Figure 2
Experimental Setup

By using a CDC camera (Samsung Digital Cam CM OS 6.4 Mega Pixel Full HD), it was made a video during about 10 seconds. From that video, it was obtained an image sequence of 512 images, using some computational routines applied in these computational programs: Matlab and ImageJ (this last one can be free downloaded in http://rsbweb.nih.gov/ij/).

After this, it was used more two computational routines in ImageJ to obtain a value of the parameter called Moment of Inertia (MI), the Modified Correlation Matrix (MOM) and the STS standard (Spatial Temporal Speckle).

With that information about moisture and MI, it was made the correlation of the two parameters to be able to do the indirect determination of moisture using the MI. It was drawn an exponential trend line.

Results and discussion

All the obtained points were used to plot the chart showed in Figure 3. Then, it was drawn an exponential trend line.

Figure 3
Relation between the moisture (%) and the observed MI to each point

It’s possible to see that when there was the increase of the moisture, the observed MI increase too. The behavior of this raise follows the y = 1,5287e^2,7857xEquation 2:

(2) $\mathit{y}\mathit{=}\mathit{1}\mathit{,}\mathit{5287}{\mathit{e}}^{\mathit{2}\mathit{,}\mathit{7857}\mathit{x}}$

The reliability of this correlation is 81,34%.

Enes et al. (2005) also verified the percentage of moisture and the MI result. The difference with this study is that they didn’t show explicitly the correlation between moisture and MI; they observed the behavior of each parameter along the time. But they observed that along the time both parameters presented reduction.

To obtain those values, it was generated some image sequence. Based on that image sequence, it was taken some steps using some computational routines, which resulted in one STS standard (Spatial Temporal Speckle) and one Modified Correlation Matrix (MOM) to each time of the water absorption analysis. Those two image results are showed at Figure 4 and Figure 6, respectively.

Figure 4
Variation of STS Standard during the 27 h of water absorption, from A to Q, respectively, 0 h; 0,25 h; 0,50 h; 0,75 h; 1,0 h; 1,5 h; 2,0 h; 2,5 h; 3,0 h; 4,0 h; 5,0 h; 6,0 h; 7,0 h; 21,0 h; 23,0 h; and 27,0 h

Figure 5
STS e MOM of low and high activity materials (Adapted by RABELO, 2000)

Figure 6
Variation of Modified Correlation Matrix during the 27 h of water absorption, from A to Q, respectively, 0 h; 0,25 h; 0,50 h; 0,75 h; 1,0 h; 1,5 h; 2,0 h; 2,5 h; 3,0 h; 4,0 h; 5,0 h; 6,0 h; 7,0 h; 21,0 h; 23,0 h; and 27,0 h

At Figure 4 it’s possible to observe a significant difference between the STS standard at the first time and the others. In the beginning of the analysis, it’s possible to see some almost straight lines, which represent a low activity material. In that case, this activity represents just the water activity because the grains were previously dried using a high temperature. That image looks like the STS standard to low activity materials, as Rabelo (2000) shown at his research (Figure 5).

Differently of this first image, while the grain was absorbing water, the MI value was growing and those almost straight lines look like interference on television. It represents a high activity material, as Rabelo (2000) showed at (Figure 5).

Analogously to STS Standard, it’s possible to observe that during the water absorption there are the increases of MI values and the variation of the Modified Correlation Matrix. Rabelo (2000) said that during the increase of MI values, there are more non-nulls values at main diagonal. This behavior can be observed at the image sequence showed at Figure 6: at time 0h it’s possible to see a diagonal without points out of the main diagonal. Along the process, a kind of cloud of points beginning to appear and the main diagonal beginning to be less defined. That’s the same behavior that could be seen at the example at Figure 4.

Conclusion

It was found the correlation between the moisture and the Moment of Inertia value of the grains: the activity grows with the water absorption.

• 1
  Parte da Tese de Doutorado do primeiro autor, apresentada ao Programa de Pós-Graduação em Engenharia Agrícola da UNICAMP, com financiamento da FAPESP (Processo número 2013/11652-9)

ACKNOWLEDGMENT

The authors would like to thank the Foundation for Supporting Research of the State of São Paulo (FAPESP) - process number 2013/11652-9.

References

• Alves, J. A.; Braga JÚNIOR, R. A.; Vilas Boas, E. V. de B. Identification of respiration rate and water activity change in fresh-cut carrots using bisopeckle laser and frequency approach. Postharvest Biology and Technology, v. 86, p. 381-386, 2013. Disponível em: http://www.sciencedirect.com/science/article/pii/S0925521413002342 Acesso em: 24 maio 2019.
  » http://www.sciencedirect.com/science/article/pii/S0925521413002342
• Amaral, I. C. et al Application of biospeckle laser technique for determining biological phenomena related to beef aging. Journal of Food Engineering, v. 119, n. 1, p. 135-139, 2013.
• Arefi, A. et al Non-destructive identification of mealy apples using biospeckle imaging. Postharvest Biology and Technology, v. 112, p. 266-276, 2016.
• BRASIL. Ministério da Agricultura, Pecuária e Abastecimento. Secretaria de Defesa Agropecuária. Regras para análise de sementes Brasilia, 2009. 399 p.
• ENES, A. M. et alBiospeckle laser em tecidos vivos e tecidos mortos de sementes de feijão (Phaseolus Vulgaris L.) durante a perda de água. In: CONGRESSO BRASILEIRO DE ENGENHARIA AGRÍCOLA, 34., 2005, Canoas, RS. Anais [...]. Canoas, 2005.
• GUEDES, J. D. S. et al Assessment of the biological activity of kefir grains by biospeckle laser technique. African Journal of Microbiology Research, v. 8, n. 27, p. 2639-2642, 2014.
• RABELO, G. F. Avaliação da aplicação do speckle dinâmico no monitoramento da qualidade da laranja. 2000. 149 p. Tese (Doutorado em Engenharia Agrícola) - Universidade Estadual de Campinas, Campinas, 2000.
• RODRIGUES, S. et al Efeito da umidade na determinação da atividade biológica de sementes de feijão (Phaseolus vulgaris L.), utilizando imagens de speckle dinâmico. Revista Brasileira de Armazenamento, v. 30, p. 135-139, 2005.

Publication Dates

• Publication in this collection
  03 Feb 2020
• Date of issue
  2020

History

• Received
  08 Apr 2016
• Accepted
  04 July 2019