تشخیص غیرمخرب بیاتی نان با استفاده از تصاویر فراطیفی

تصویربرداری فراطیفی، ترکیبی از فناوری تصویربرداری و طیف‌سنجی است که مقادیر زیادی از اطلاعات فضایی و طیفی را به‌طور همزمان ارائه می‌دهد، و امروزه به‌عنوان یک ابزار تشخیص غیرمخرب و سریع برای ارزیابی کیفیت و ایمنی مواد غذایی در حال گسترش است. در این پژوهش با استفاده از تصویربرداری فراطیفی، در محدوده طول موج nm400-950 و با وضوح کیفی nm 0.795، چگونگی فرایند بیات شدن نان و تاثیر آن بر رفتار نان بررسی شد. بعد از استخراج مولفه‌های اصلی، به منظور پیش‌بینی ویژگی‌های بافتی از سه روش مدل‌سازی pcr، plsr و grnn طی شش روز نگهداری استفاده شد؛ نتایج نشان دادند روش grnn نسبت به دو روش دیگر دارای بیشترین مقادیر ضریب تشخیص r2 برای دو ویژگی، ارتجاعیت و سفتی به ترتیب 0.96 و 0.94 و همچنین کمترین مقدار خطا rmse برای دو ویژگی، پیوستگی و سفتی به ترتیب 0.11 و 0.32 می‌باشد که نشاندهنده توانایی مدل شبکه عصبی رگرسیون تعمیم‌یافته برای پیش‌بینی ویژگی‌های بافتی نان است.

non-destructive detection of bread staleness using hyperspectral images

introduction: bread is a crucial source of sustenance, yet its texture is subject to rapid deterioration through a process commonly referred to as staling. the development of a non-destructive method for the expeditious evaluation of textural changes in bread during storage would facilitate research of the impact of various additives on bread’s quality attributes and shelf life. given that bread is among the most perishable of processed foods, devising techniques for the prompt detection of staling is of paramount importance. in recent years, hyperspectral imaging (hsi) systems have emerged as a highly precise and accurate non-destructive means of assessment and diagnosis within the agricultural and food industries. hsi cameras enable the scanning and observation of concurrent chemical mechanisms within bread’s texture. hsi, also known as chemical or spectroscopic imaging, is an analytical technique that amalgamates chemical data derived from spectroscopy with spatial information pertaining to the surface under examination. as such, a sample’s hyperspectral image contains an abundance of spatial and spectral-chemical data that are typically highly intercorrelated. the objective of this study was to assess bread staling through the utilization of hyperspectral imaging.materials and methods: to produce baguette bread dough, wheat flour was combined with sugar (1%), salt (1%), yeast (1.5%), and improver (1.5%) by weight of flour, along with the requisite quantity of water. the dough was then baked at 250℃ for 15 minutes. this study aimed to investigate the staling process of baguette bread by examining changes in the physical and sensory characteristics, as well as hyperspectral images, of samples stored at room temperature up to six days. sample weights were recorded on days zero, two, four and six to determine the moisture content of the crust and crumb. the texture profile analysis (tpa) test was employed to assess sample texture during storage, with firmness, chewiness, cohesiveness, and springiness characteristics being calculated. hyperspectral images were captured using a linear system operating within the 400-950 nm wavelength range with a spectral resolution of 0.795 nm. hyperspectral image preprocessing was then conducted to eliminate spatial and spectral noises. following preprocessing, principal component regression (pcr), partial least squares regression (plsr) a generalized regression neural network model were utilized to predict texture characteristics. all procedures were executed using matlab 2022b and sas software.results and discussion: analysis of bread texture revealed that, over the course of six days of storage, the mass and moisture of the crumb exhibited a significant decreasing trend at the 5% probability level. conversely, a significant increase in crumb moisture and firmness was observed during storage (p<0.05). springiness also demonstrated a significant decreasing trend, in contrast to cohesiveness. to ascertain the number of principal components containing the majority of image information, a graph depicting the number of principal components relative to the percentage of variance in each component was plotted cumulatively. it was determined that, for both pcr and pls methods, the percentage variance in three principal components exceeded 96%. with increasing bread storage duration, the percentage of positive components for all three principal components increased. for pc1, pc2, and pc3 on first day, these characteristics were 57%, 41%, and 28%, respectively, while on the final day of storage they were 98%, 61%, and 57%, respectively. this indicated a significant increasing trend at the 5% probability level for all three principal components during storage. these findings are entirely consistent with texture profile analysis (tpa) measurements.