Correlation between rheological properties and maturity of passion fruit based on machine vision

Rheological properties play an important role in food production and quality control. This research explores the relationship between rheological parameters and quality characteristics of passion fruit and establishes a maturity classification model for passion fruit based on its rheological properties. Each sample undergoes a rheological test, texture profile test, puncture test, and physicochemical index test. These tests aim to gather precise mechanical and physiological information on passion fruit. We built a mechanical testing platform and used machine vision to analyse the micro-deformation of fruit. The platform can measure the real-time contact area and load value to obtain accurate stress values during compression. Non-destructive rheological tests were conducted on intact passion fruit to get the elastic modulus during the loading stage. It is highly consistent with the results of traditional Hertz contact theory. Additionally, the stress relaxation parameters were obtained by fitting the five elements Maxwell model during the holding stage. Notably, there are strong correlations between the rheological parameters and most texture parameters or physicochemical indicators, with the highest correlation coefficient reaching 0.703. Therefore, the rheological parameters were utilised as inputs for maturity classification models (GBDT, MLP, and AdaBoost). All models achieved satisfactory classification results. Particularly, the GBDT model demonstrated excellent classification performance and generalisation capability, with Precision, Recall, and F-Score of 80.44%, 80.08%, and 80.26%. The results show that it is feasible to determine the maturity of passion fruit based on non-destructive rheological characteristics.