Researches regarding the image analysis in wheat quality evaluation

Wheat grain color can vary from light yellow to brown red. The main factors which determine the color  are some flavonoid pigments (tricine) and carotenoids (xanthophylls such as lutein).

The analysis of color based on digital images of wheat grains has  been used in various studies. Neuman et al. (1989) have shown that the method can be used to distinguish certain varieties of Canadian wheat [1].

Klepacka et al. (2002) have shown the existence of a significant relationship between the wheat bran shades of gray of and ferulic acid content of these (r = -0.65). They also found a correlation between the level of gray color grains and the degree of extraction (r = 0.74) [2].

Manickavasagan et al. (2008) studied the potential of the wheat image analysis (using monochrome images based on the shades of gray) to provide information which allow the discrimination of different classes of wheat, to automation of industrial processes [3].

Newton’s experiments and confirmed by the Young-Helmholtz theory, demonstrated that the human eye retina contains three types of cone receptors, each being sensitive to a certain range of light waves. These receptors are: Long or Red receptors (sensitive to red light with long wavelengths, in the range 500 nm-700nm), the Green or Middle receptors (sensitive to green light having medium wavelengths, 450nm-630nm) and Short or Blue receptors (sensitive to blue light with short wavelengths, in the range 400nm- 500 nm) [4,5].

In practice, the description of any color in the visible spectrum consists of its noting, representation or specification, through three numerical color parameters, which define a set of tristimulus values. A tristimulus value expresses, directly or indirectly, the extent to which primary RGB colors combine to form a new color. Implicitly, it expresses the characteristics of color stimuli, which are sensitive to LMS wavelengths, corresponding to the primary color components (RGB).

The scanners read the amounts of light reflected by a RGB image and convert them in the tristimulus values (digital), and monitors receive tristimulus values (digital) and convert them in RGB light, visible on the fluorescent screen.

The RGB color model can be implemented in different ways. The range of colors which can be described using this model is determined, dimensionally, by the number of bits used to describe each color component. The most common mode of implementation, used since 2006 for computer monitors, uses 24-bit color and 8 bit color / pixel or 256 digital levels / channel (28 = 256), which is why the number of colors that can be represented based on this model is limited to 256R x 256G x256B = 16.7 million colors, about the number that can be distingushed by human eye [5].

Since several studies have shown significant correlations between the levels of certain compounds in wheat grain and its color, we decided to investigate the relationship between the color of wheat grains and their technological quality, in the Romanian wheat samples .

There have been analyzed the main quality parameters of 27 wheat samples from Romanian crops, of the years 2010 and 2011, namely: Humidity (%), Hectolitric mass (kg / hl), Falling number (sec), Protein content (%), Wet gluten content (%) and Gluten Index. Afterwards, the wheat samples were ground on the Chopen pilot-type mill and the alveografic parameters were determined from the resulting flour: Resistance (P, mm), Extensibility (L, mm), Mechanical Work (W, 10E-4J), Elasticity index (Ie,%), Gluten Extensibility index and the P/L report.

The wheat samples taken for analysis were scanned using a commercial scanner, at a resolution of 200 dpi. The obtained results were examined with a specific software, used for the analysis of ImageJ image. For each image we determined the color histogram with the specific parameters: R, G, B, Brightness and Fractal Dimension.

Our results have shown highly significant correlations between the color histogram parameters (Brightness, R, G, B) and the alveografic parameters, namely: Resistance (-0.64< r < – 0.72), Mechanical work (- .62 < r < -0.64) and the P/L ratio (-0.62 <  r < -0.68). No significant correlations were found between color parameters of the image analysis and the physical and chemical parameters of the wheat.

Models of multiple regression have been described for the prediction of alveografic parameters P, W and P/L, based on the color parameters. The results suggest an interesting potential for including the image analysis in a coherent assessment procedure of the wheat, but in order to validate this conclusion, there are necessary further experimental researches.

Full paper: art71

REFERENCES

[1] Neuman M.R., Sapirstein H.D.,  Shwedyk E., Bushuk W., 1989, Wheat grain colouranalysis by digital image processing II.Wheat class discrimination, Journal of Cereal Science Volume 10, Issue 3, November 1989, Pages 183–188

[2] Klepacka J., Fornal Ł., Konopka S., Choszcz D., 2002, Relations between ferulic acid content in wheat coat, and milling quality and colour of grain, Electronic Journal of Polish Agricultural Universities, 2002, Volume 5, Issue 2, Series Food Science and Technology

[3] Manickavasagan A., Sathya G., Jayas D.S., White N.D.G., 2008, Wheat class identification using monochrome images, Journal of Cereal Science, Volume 47, Issue 3, May 2008, Pages 518–527

[4] Nemeth R., 2009, Management of Color – Specialized courses, Technical College, Constantin Brancusi, Oradea http://cobra.rdsor.ro/.

[5] Tamba – Berehoiu S.M., Popa N.C., Tamba – Berehoiu R., Popescu R., Culea R.,, 2010, Investigations on the use of color in the marketing of milling products, Scientific Papers “Management, Economic Engineering in Agriculture and Rural Development”, Volume 10(1)2010, ISSN 1844- 5640

[6] Thierer L. Volf, 1971, Determinarea calităţii produselor agricole vegetale, Ed. Ceres, Bucureşti, pg. 121