Evaluation of Image Analysis Tools for Characterization of Sweet Bread Crumb Structure

Abstract

Many approaches to evaluate bread crumb features by applying free or at least not too expensive image analysis (IA) software have been published; however, the described procedures showed noticeable differences. The aim of this work was to compare different image scanning resolutions and thresholding techniques to quantify sweet bread crumb features (cell density, mean cell area, shape factor) and their relation with fractal dimension. Two sets of experiments were carried out, one to determine the effect of scanning resolution and thersholding method and the other to validate the previous results by evaluating breads with different crumb structures. Nine different scanning resolutions (75, 100, 150, 200, 300, 355, 435, 515, 555 dpi) and two segmentation procedures (Otsu and Manual) were tested. Three different types of commercial sweet breads and a yeasted sweet bread added with different concentrations (six, 12%) of Chia flour (Salvia hispanica) were evaluated. Results showed that the percentage of particles with areas between 0.1 and 4.0 mm 2 remained almost constant when using 350 dpi or larger resolution values, while the smallest particles (<0.1 mm 2 ) increased their proportion up to 87% at the highest scanning resolution for both thresholding methods. IA was useful to detect crumb structure differences among commercial breads and breads added with Chia flour as obtained from cell density (154 ± 4.6-246 ± 2.5) and mean cell area (0.81 ± 0.02-0.7 ± 0.03) results. However, the number of selected objects to calculate these parameters produced different results. The addition of 6% of Chia flour did not affect the bread crumb features, while at the largest proportion more and smaller pores were obtained. Fractal texture was useful to evaluate bread crumb structure, as it not depends on the number of particles detected.

Optimization of image analysis techniques for quality assessment of whole-wheat breads made with fat replacer

“…Both with Dtsu's method and with the default ImageJ algorithm (Table 2), the highest cell density and lowest mean cell area values were found with the highest whole-wheat flour content, which is consistent with the results obtained by Farrera-Rebollo et al (2012) in breads made with chia flour and by Hruskova et al (2012) in whole-grain breads.…”

“…Although a standard procedure does not yet exist, several studies have used image analysis to assess bread loaves made with refined wheat flour (Zghal et al, 1999;Gonzales-Barron & Butler, 2006) and some made using whole grains, as soy (Lodi & Vodovotz, 2008), β-glucan (Skendi et al, 2010), chia (Farrera-Rebollo et al, 2012), amaranth (Sanz-Penella et al, 2013) and whole wheat (Torri et al, 2013).…”

An integrated technological approach to the selection of lactic acid bacteria of flour origin for sourdough production

“…Recently, image analysis has been used as a quantitative tool for the assessment of crumb features (Farrera-Rebollo et al, 2012). In this study, the highest values of void fraction and cell density were registered for all W. cibaria strains, Ln.…”

Image Processing Applied to Classification of Avocado Variety Hass (Persea americana Mill.) During the Ripening Process

“…Fractal dimensions (FD SDBC ) was estimated by the Shifting Differential Box Counting method from GL images (Mapfractalcount plug-in version 1.0, ImageJ). Textural features and FD SDBC meanings have been described previously (Haralick et al 1973;Mendoza et al 2007;Quevedo et al 2008;Farrera-Rebollo et al 2011). Results are reported as averages and standard errors.…”