Statistical pattern recognition classification with computer vision images for assessing the furan content of fried dough pieces

Abstract

This research tested furan classification models in fried matrices based on the pattern recognition of images. Samples were fried at 150, 160, 170, 180, and 190 °C for 5, 7, 9, 11, 13, and 30 min. Furan was measured by GC–MS. Corresponding images were acquired and processed to extract 2175 chromatic and textural features. Principal component analysis was used to reduce features to 8–12 principal components. In parallel, sequential forward selection coupled with linear discriminant analysis (LDA) was the best strategy to select only 5–7 features. LDA was the best classifier with 91.39–97.60% recognizing above 113 µg/kg and 69.54–83.80% to classify images from class 1 (0–38 µg/kg) from class 2 (39–113 µg/kg). Also, support vector machine recognized 87.71–96.74% of class 3 (114–398 µg/kg) from class 4 (399–646 µg/kg). The technique may be used to detect high amount of furan in fried starchy matrices.