Detection of browning of fresh-cut potato chips based on machine vision and electronic nose

This study explored the feasibility of achieving rapid nondestructive detection of browning in potatoes from a visual and olfactory perspective. In five grades, 15 pieces were taken out from each batch, respectively, and tested for browning characteristics using Chroma meter (CM), machine vision (MV), and electronic nose (E-nose). Linear discriminant analysis, support vector machine (SVM), K-Nearest Neighbor, and Back Propagation Artificial Neuron Network were used to classify the samples. The discriminant accuracy of the MV and E-nose datasets was 0.960 and 0.813, respectively, and this increased to 1,000 after data fusion. Partial least squares regression and SVM regression were also used to investigate the correlation between two data sets and CM data. The MV results for L*, a*, and b* were 0.864, 0.966, and 0.992, with variances of 1.062, 0.575, and 0.123, respectively, indicating a very strong correlation. The predicted result of E-nose for a* reached 0.705, and the variance was 1.815, which also showed a high correlation. The data fusion models for L*, a*, and b* yielded 0.883, 0.968, and 0.997, with variances of 1.264, 0.633, and 0.083, respectively. These results indicated that MV and E-nose can be used as nondestructive detection tools for detecting browning in potatoes. Practical Applications Potato products are popular among consumers because of its rich nutrition and processing characteristics. However, enzymatic reactions during peeling and processing of fresh-cut potato results in color defects which compromise the quality of the final products. Therefore, it is an inevitable requirement for the development of the industry to adopt appropriate methods to detect and prevent browning in fresh-cut potatoes as a means of strengthening quality management. The present study explored the feasibility of rapid nondestructive detection of browning in potato from the visual and olfactory aspects. A new idea for the application of machine vision and electronic nose in the field of nondestructive detection of agricultural products was proposed. At the same time, this article has certain reference value to the quality management of potato and other fruit and vegetable processing industry.