Fruit and vegetable quality detection method based on multispectral image information and TLMD-WOA-SIFT

The invention relates to a fruit and vegetable quality detection method based on multispectral image information and TLMD-WOA-SIFT. The fruit and vegetable quality detection method comprises the following steps: (1) acquiring multispectral image information of fruits and vegetables; (2) decomposing images by using two-dimensional local mean decomposition (TLMD) to generate a plurality of two-dimensional product functions (BPF); (3) extracting features of the BPF by using a WOA optimized SIFT (scale invariant feature transform) algorithm; (4) processing feature information extracted by the BPFto obtain all feature information of an original multispectral image; and (5) inputting the feature information into an LSSVM model for classification. TLMD is combined with WOASIFT for fruit and vegetable quality detection, so that the invention has very high recognition accuracy and practical value, is high in generalization ability, and is very suitable for real-time accurate detection and popularization of the fruit a