Objective quality assessment of apples using machine vision, NIR spectrophotometer, and electronic nose.

Three different sensors, a near-infrared spectrophotometer (NIR), a machine vision system (MV), and an electronic nose system (EN), were combined for non-destructive quality detection of 'Fuji' apples. The intention was to take advantage of the three sensors, one performing a local measurement of one physical property of the fruit (sugar content) and the others performing a global assessment of other physical properties (color, size, shape, and aroma), and combine those types of measurement (local and global) to improve the accuracy of quality assessment. The EN was used to assess the rotting stage of apples based on ANN (artificial neural network). A relationship was also found between sugar content and different NIR wavelengths by using MLR (multiple linear regression). The surface color, shape, and size of apples were assessed by MV technique. The three sensors were working at the same time. A total of 104 'Fuji' apples were detected by the three-sensor combination system and were divided into two sets, with 84 in set A and 20 in set B. By combining the three different kinds of sensors, it is shown that the accuracy of quality assessment of apples can be improved with a high-level fusion technique. For sugar content assessment, the classification error rate dropped from around 17% using only NIR spectra to around 6% when the three sensors were combined through ANN. Finally, the three sensors were combined to evaluate the quality of apples through a decision tree, and only six apples in set A and one apple in set B were misclassified. The results indicate that the three-sensor combination has a higher accuracy for classification and is promising for both customers and producers in assessing the quality of apples.