The mango core quality non-destructive testing and comprehensive harvesting decision-making model based on portable near-infrared spectroscopy technology.
- 2026-03-20
- Frontiers in plant science 17
- PubMed: 41937797
- DOI: 10.3389/fpls.2026.1772717
Study Design
- Population
- three mango varieties: 'Tainong', 'Guifei', and 'Jinhuang'
- Methods
- Collected absorption spectra between 650-950 nm at 20 days before and after commercial harvesting; measured fruit firmness, pH, SSC, DMC; used Kennard-Stone algorithm and six preprocessing methods to build models
- Duration
- 20 days before and after commercial harvesting
It is of great significance to detect the degree of mango maturity accurately and quickly in terms of timely harvesting according to market needs and building a differentiated post-harvest field sorting system. In this study, three main mango varieties-'Tainong', 'Guifei', and 'Jinhuang'-that originated in Hainan, China, were used as subjects. An H-100F portable near-infrared spectrometer was used to collect the absorption spectra between 650- and 950-nm wavelengths at 20 days before as well as 20 days after the commercial harvesting time. Four types of quality indicators, including fruit firmness, pH value, soluble solid content (SSC), and dry matter content (DMC), were measured in the laboratory to finally establish the non-destructive detection (NDT) and comprehensive harvesting decision-making model. As the results indicated, in the range of 650-950 nm, the average value of the spectrum absorbance for the three mango varieties decreased as the grade of maturity increased. The value change correlated significantly with fruit firmness, pH, SSC, and DMC (%). The Kennard-Stone algorithm method was used to divide the four quality indicators into correction sets and prediction sets. The original spectrum and the spectrum preprocessed using six methods, including multiplicative scatter correction, standard normal variate transform, Savitzky-Golay convolution smoothing, Savitzky-Golay convolution derivative, vector normalization, and maximum-minimum normalization, were then preferably selected to build a processing model for the four quality indicators of the three mango varieties. The test results showed that for all three mango varieties, both the revised standard error of mean squared root and the indicated standard error of mean squared root of the four quality indicators were less than 1.42, 0.62, 0.78, and 1.4, and 1.52, 0.38, 0.94, and 1.5, respectively. The accuracy rates of the model to test the grade of maturity for 'Tainong', 'Guifei', and 'Jinhuang' were 83%, 90%, and 81%, respectively. The above results indicated that the mango NDT model that was established on the portable near-infrared spectroscopy technology can reliably detect the fruit firmness, pH, SSC, and DMC of mangoes and can distinguish the grade of maturity for different mango varieties. The research results are of great significance to the decision-making of mango harvesting and the differentiated field sorting thereafter.
Research Insights
| Supplement | Dose | Health Outcome | Effect Type | Effect Size | Source |
|---|