RESUMO
Objective: With the improvement of living standards, consumers are paying more and more attention to the quality of rice. Traditional rice quality detection relies on human sensory judgment, which is inaccurate and inefficient. With the continuous development of molecular imaging technology, more and more scholars at home and abroad have begun to pay attention to its application in the nondestructive testing of agricultural products. Molecular imaging technology combines the advantages of spectral technology and image technology, which can achieve rapid, nondestructive and accurate detection of rice quality. In this paper, taking rice as the research object, we carried out nondestructive detection research on rice varieties, moisture and starch content using molecular imaging technology. We proposed a rapid detection method based on molecular imaging technology for rice variety identification, moisture content and starch content. Molecular images of the rice samples from four origins were obtained using a molecular imaging system, the regions of interest of the rice were identified and, spectral data, textural features and morphological features of the rice were extracted. Spectral, textural and morphological features were selected by principal component analysis (PCA), and nine feature wavelengths were obtained and an optimal model was established with an accuracy of 91.67%, which demonstrated the feasibility of molecular imaging. By comparing the models, the BCC-LS-SVR model based on the RB function had the highest accuracy with R2 of 0.989, RMSEP of 0.767%, R2 of 0.985, and RMSEC of 0.591%. Moreover, starchy rice was detected using molecular imaging. The PCA-SVR model based on the RBF kernel function had the highest accuracy with R2 of 0.989, RMSEC of 0.445%, R2 of 0.991, and RMSEP of 0.669%. Our models demonstrated high accuracy in identifying rice varieties, as well as quantifying moisture and starch content, showcasing the feasibility of molecular imaging technology in rice quality assessment. This research offers a rapid, nondestructive, and accurate method for rice quality assessment, promising significant benefits for agricultural producers and consumers.