外文摘要：The use of innovative techniques in the classification of dried products for food processing and packaging systems simplifies and brings objectivity to the related processes. In addition, it helps consumers to access quality products. This study aimed to discriminate dried carrot samples subjected to 12 different combinations of drying and pretreatment by using innovative techniques based on chromatic attributes. The drying procedure consists of three different microwave powers (100, 200, 300 W), and pretreatments include 5% gum arabic solution, 5% gum sucrose solution, ultrasound application, and a control (no -pretreatment). The R, G, B, L, a, b, H, S, and V color channels were extracted from dried carrot images, and the exploration of classification models from selected color channels was an innovative aspect of the study. In all color channels, high overall accuracies were obtained with the values of 96.33, 96.50, and 97.00% for the logistic model tree, k -nearest neighbor, and random forest, respectively. The confusion matrix showed the lowest mixing among carrots dried at H -US (300 W + ultrasound) vs. H -C (300 W + control), H -S (300 W + sucrose) vs. H -C (300 W + control), and M -S (200 W + sucrose) vs. H -S (300 W + sucrose). The highest L, a, and b color values for the dried carrot slices were obtained as M -S (75.26), H -GA (30.93), and M -GA (44.75), respectively. In addition, hue values varied between 17.26 and 13.27, whereas saturation values varied between 50.93 and 59.90. The finding revealed that assessment of the effect of pretreatment and drying on carrot slices could be performed by machine learning and image analysis in a practical, rapid, and non-destructive manner.
外文关键词：machine learning;image analysis;discrimination;Pretreatment;Dried carrot slices
作者：Ropelewska, Ewa;Cetin, Necati;Gunaydin, Seda;Sacilik, Kamil
作者单位：Ankara Univ;Natl Inst Hort Res;Univ Erciyes
期刊名称：JOURNAL OF FOOD COMPOSITION AND ANALYSIS
期刊影响因子：0.0
出版年份：2024
出版刊次：129
原文传递申请：江苏省科技资源（工程技术文献）统筹服务平台