Prediction model of soluble solids content in fresh peach based on hyperspectral image

In this study, a 400-1000nm hyperspectral camera is used, and the product FS13 of Hangzhou CHNSpec technology Co., Ltd. can be used for related research.

Fresh peach is a kind of fruit with rich nutrition and sweet flavor, Soluble solid content (SSC), as an important component affecting the flavor of fresh peach, has also become an important reference standard to measure the quality of fresh peach. Therefore, accurate estimation of SSC has important research significance and application value for fresh peach grading and evaluation. At present, with the rapid development of sensor and data analysis technology, nondestructive estimation of fruit soluble solid content has been widely studied and applied. Among them, near infrared spectroscopy SSC in fresh fruit has been successfully detected by spectrum, multispectral, fluorescence spectrum and electronic nose. However, most studies are based on single feature detection, which limits the further exploration of fruit SSC prediction model. In recent years, hyperspectral images provide not only spectral dimension information, but also spatial dimension information, which is often widely used to detect SSC of fruits. The results show that it is feasible to estimate SSC based on hyperspectral image features. However, most studies are only based on spectral dimension information, which is easy to lead to over fitting of SSC estimation model. With the application of deep learning in different fields, it provides a new idea and scheme for SSC prediction of fresh peach. As a deep learning method, stacked automatic coder (SAE) has strong feature ability to improve the accuracy of prediction model. Therefore, in this study, stacked automatic coders with different structures are designed to extract the deep features of spectral dimension and spatial dimension information of hyperspectral images respectively, so as to provide a technical path for the quantitative analysis of fresh peach SSc.

The SSC visualization of fresh peach samples of different varieties shows that sae-pso-svr model has good universality. The deep features of spectral information and spatial information of fresh peach hyperspectral image are extracted based on SAE, and the fresh peach SSC estimation model based on sae-pso-svr is constructed by fusing the deep features of information, which effectively improves the estimation accuracy of the model.