The invention discloses a method for multispectral nondestructive detection of fruit and vegetable
sugar degree, acidity and maturity, and relates to the technical field of fruit and vegetable detection, historical sample data is collected, candidate
wave band combination is established, according to an analysis result, collection configuration is set, real-time collection data is formed,
dark current deduction,
gain unification, polarization mirror inhibition and
visual angle normalization are executed, and the fruit and vegetable
sugar degree, acidity and maturity are detected. According to the method, the
sugar degree, acidity and maturity of
fruits and vegetables can be efficiently and nondestructively detected, the sample error and sampling time in traditional detection are reduced, the
data quality is improved, accurate fruit surface
information extraction is ensured, and the accuracy of fruit surface detection is improved. The accuracy of sugar degree, acidity and maturity prediction is remarkably improved through multi-dimensional
feature extraction and multi-task modeling, the method is more stable especially in a complex environment, in addition, efficient decision support is provided for agricultural production, an accurate harvesting time window and a sorting strategy are generated, resource waste is reduced, and the production efficiency is improved.