[Determination of Soluble Solid Content in Strawberry Using Hyperspectral Imaging Combined with Feature Extraction
Guang Pu Xue Yu Guang Pu Fen Xi = Guang Pu
|July 23, 2015
Summary
Hyperspectral imaging accurately determines soluble sugar content (SSC) in strawberries. This non-destructive method utilizes feature extraction techniques for precise quality assessment in fruit.
Area of Science:
- Agricultural Science
- Food Science
- Spectroscopy
Background:
- Accurate determination of soluble sugar content (SSC) is crucial for strawberry quality assessment.
- Traditional methods for SSC measurement can be destructive and time-consuming.
- Non-destructive techniques like hyperspectral imaging offer a promising alternative for rapid quality evaluation.
Purpose of the Study:
- To investigate the application of hyperspectral imaging combined with feature extraction for determining SSC in strawberries.
- To compare the performance of different feature extraction methods in developing predictive models for SSC.
- To establish a reliable and non-destructive method for SSC quantification in strawberries.
Main Methods:
- Hyperspectral images of strawberries were acquired in the 874-1,734 nm spectral range.
- Spectral data preprocessing included moving average (MA) filtering.
- Feature extraction methods such as Successive Projections Algorithm (SPA), Genetic Algorithm Partial Least Squares (GAPLS), Weighted Regression Coefficient (Bw), Competitive Adaptive Reweighted Sampling (CARS), Principal Component Analysis (PCA), and Wavelet Transform (WT) were employed.
- Partial Least Squares (PLS) regression models were built using full spectra, selected wavelengths, and extracted features.
Main Results:
- PLS models demonstrated good performance in predicting SSC.
- Models utilizing full spectra and features extracted by Wavelet Transform (WT) achieved the highest accuracy.
- Correlation coefficients for calibration (r(c)) and prediction (r(p)) exceeded 0.9 for the best models.
- Feature extraction methods effectively reduced data dimensionality while preserving crucial spectral information.
Conclusions:
- Hyperspectral imaging, coupled with appropriate feature extraction techniques, is a viable non-destructive method for determining SSC in strawberries.
- Wavelet Transform (WT) proved effective in extracting features for robust SSC prediction.
- This approach offers potential for real-time quality control and sorting of strawberries based on sugar content.
