The storage time of lettuce is an important factor affecting the quality and nutritional value of lettuce. In order to identify the storage time of lettuce quickly, effectively and non-destructively, Adaboost algorithm was used to identify and classify lettuce by near infrared spectroscopy. Taking 100 fresh lettuce samples as the research object, the near infrared diffuse reflectance spectrum of lettuce was detected every 12 hours by Antaris II near infrared spectrum analyzer for three times, and the spectrum scanning wave number ranged from 10000-4000cm-1.Firstly, the standard orthogonal transform (SNV) is used to preprocess the collected data to eliminate the influence of noise, and then the principal component analysis (PCA) is used to reduce the dimension of the 1557-dimensional lettuce near infrared spectrum data. After PCA, the spectral data information after dimension reduction is extracted by LDA algorithm, which improves the accuracy of clustering and achieves the maximum distance between classes and the minimum distance within classes. Then, the weak classifier in Adaboost algorithm is constructed by using K-nearest neighbor rule.After 10 iterations, the strong classifier composed of 10 K-nearest neighbor weak classifiers can achieve classification accuracy higher than 98%, and the classification time is about 3.6s.Experiments show that Adaboost technology combined with dimensionality reduction clustering algorithms such as PCA and LDA provides a new idea for quickly and efficiently identifying the storage time of lettuce.

Mao Xinyuan. Design of storage time identification system of lettuce based on Adaboost. Academic Journal of Engineering and Technology Science (2020) Vol. 3 Issue 7: 201-209. https://doi.org/10.25236/AJETS.2020.030720.

[1] Hu Yue, Cui Wen, Jin Minfeng, et al. Effects of different cultivation methods on the growth and nutritional quality of lettuce [J]. Journal of Shanghai Normal University (Natural Science Edition), 2019,48(05):566-573.
[2] Parrini S, Acciaioli A, Franci O, et al. Near Infrared Spectroscopy technology for prediction of chemical composition of natural fresh pastures[J]. Journal of applied animal research, 2019, 47(1): 514-520.
[3] Yin Shikui, Feng Guohong, Li Chunxu, et al. Optimization of coniferous wood basic density estimation model based on near infrared spectral band optimization [J]. Journal of Central South University of Forestry and Technology, 2020, (03):85-95.
[4] Cheng Jiabing, Zou Xiangjun, Lin Guichao, et al. Fast detection method of green litchi by cascade classifier based on AdaBoost algorithm [J]. Automation and Information Engineering, 2018,39(05):38-44.
[5] Anna K G E, Amir H E, Andres E. Classification models for heart disease prediction using feature selection and PCA[J]. Elsevier ltd, 2020, 19: 67-74.
[6] Tatiana V N, João C M. Barreira, et al. Phylogenetic insights on the isoflavone profile variations in Fabaceae spp: Assessment through PCA and LDA [J]. Food research international, 2015, 76:51-57.
[7] Li W, Chen Y M, Song Y P. Boosted K-nearest neighbor classifiers based on fuzzy granules[J]. Knowledge-Based systems, 2020, 195: 78-90.
[8] Xu Lihua, Xie Deti. Response of prediction accuracy of soil organic matter content to spectral pretreatment and characteristic band [J]. jiangsu journal of agricultural sciences, 2019(06):1340-1345.