Location: Sugarbeet and Bean Research
Title: Optimal Wavelengths Selection Using Hierarchical Evolutionary Algorithm for Prediction of Firmness and Soluble Solids Content in Apples Authors
|Min, Huang -|
Submitted to: ASABE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: April 28, 2009
Publication Date: June 21, 2009
Citation: Min, H., Lu, R. 2009. Optimal Wavelengths Selection Using Hierarchical Evolutionary Algorithm for Prediction of Firmness and Soluble Solids Content in Apples. Proceedings of the ASABE Annual International Meeting. Paper No. 097428. Interpretive Summary: Spectral scattering is a new technique for characterization of light scattering and absorption features in fruit for the visible and shortwave near-infrared region. It has demonstrated superior capability for prediction of fruit firmness compared to other nondestructive techniques (e.g., near-infrared spectroscopy) that have been developed previously. This research used an artificial intelligence method, called hierarchical evolutionary algorithm (HEA), to select optimal wavelengths from the hyper-spectral scattering data obtained from 'Golden Delicious' apples for predicting fruit firmness and soluble solids content. A set of optimal wavelengths was identified for firmness and soluble solids content prediction, and the HEA-based calibration models resulted in better prediction results compared with the full-spectrum partial least squares method, which is widely used in analyzing spectral data. This artificial intelligence method provides an effective means for analyzing spectral scattering data to predict fruit firmness and soluble solids content. The method can be integrated with a spectral scattering system to achieve the goal of sorting and grading apples for internal quality.
Technical Abstract: Hyperspectral scattering is a promising technique for rapid and noninvasive measurement of multiple quality attributes of apple fruit. A hierarchical evolutionary algorithm (HEA) approach, in combination with subspace decomposition and partial least squares (PLS) regression, was proposed to select optimal wavelengths from the hyperspectral scattering profiles of ‘Golden Delicious’ apples for the development of calibration models to predict fruit firmness and soluble solids content (SSC). Seventeen optimal wavelengths were selected for firmness,which nearly spanned the entire spectral range of 500 -1,000 nm, while 16 optimal wavelengths, none of them below 600 nm, were selected in the SSC prediction model. The model using 17 optimal wavelengths for predicting firmness yielded better results (r = 0.857, root mean square error of prediction or RMSEP = 6.2 N) than the full spectrum PLS model (r = 0.848, RMSEP = 6.4 N). For predicting SSC, the model using 16 optimal wavelengths model also yielded better results (r = 0.822, RMSEP = 0.78%) than the full spectrum model (r = 0.802, RMSEP = 0.83%). The proposed HEA approach provided an effective means for optimal wavelengths selection and improved the prediction of apple firmness and SSC compared to the approach of using full spectrum.