Abstract: The Raman spectra of apple skin tissue at different stages of infection by spoilage fungi were collected, and the Raman peaks were analyzed by comparing them with the Raman spectra of authentic standards. The characteristic spectral peaks of cellulose, pectin and polysaccharide were selected to construct pseudo color images showing the distribution of these three components in the cell wall and intercellular space of apples, and principal component analysis (PCA) combined with linear discriminant analysis (LDA) was used to establish a discriminant model for determining different stages of infection of five dominant spoilage fungi in apple tissue. It was found that apples had response peaks at 1 645 and 2 946 cm-1, which were assigned to cellulose, pectin and lignin. The pseudo color images showed that these components were unevenly distributed in the cell wall and intercellular space of apples. With the aggravation of the infection degree of spoilage fungi, the intensity of the characteristic Raman peaks showed a downward trend. PCA showed that the Raman spectra of apples at different stages of infection by spoilage fungi had clustering trend, and the recognition accuracy of the discriminant model for the calibration set and prediction set was more than 95%. The above results show that Raman chemical imaging can effectively characterize the infection process of apples by spoilage fungi.

Key words: apple; micro-Raman spectroscopic imaging; dominant spoilage fungi; infection process; pattern recognition