Submitted to: ASAE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: 7/1/2004
Publication Date: 8/4/2004
Citation: Peng, Y., Lu, R. 2004. Predicting apple fruit firmness by multispectral scattering profiles. ASAE Annual International Meeting. Paper No. 043007.
Interpretive Summary: Currently, fruit are sorted for size and color, but not for internal quality. Firmness is an important textural attribute that is directly related to the eating quality of apple fruit. Technologies that can sort fruit for firmness would allow the industry to deliver better quality, more consistent products to the marketplace, improve consumer acceptance and satisfaction, and increase industry profitability. This paper reports on results from a study of using an imaging technique to acquire light scattering information from apples at multiple wavelengths for predicting fruit firmness. Specifically, this research proposed a mathematical model to describe light scattering profiles for apple fruit at selected wavelengths and related scattering model parameters to apple fruit firmness. Results showed that the mathematical model described the scattering profiles accurately and was able to predict apple fruit firmness with the correlation coefficient of 0.82. The mathematical model will be incorporated into our sensing system for measuring apple fruit firmness. This research represents an important step in the development of a prototype multispectral imaging system for grading and sorting apple fruit for internal quality. The sensing technology will enable the fruit industry to separate inferior fruit from premium quality fruit, direct fruit of different quality grades for better uses, and thus enhance industry competitiveness.
Technical Abstract: Firmness is an important internal quality attribute of fruits. Development of nondestructive technology for fruit firmness measurement will help the industry provide better fruit for the consumer. This research proposed a mathematical model to describe the relationship between profile parameters of near-infrared scattering images of apples and fruit firmness. A multispectral imaging system with four filters was used for image data acquisition. Scattering images of each apple were radially described by a Lorentzian distribution function with three independent profile parameters, i.e. asymptotic value (a1), peak value (a2), and scattering width (a3) which is the full width at half maximum (FWHM). Nonlinear regression analysis was performed to obtain the three Lorentzian parameters for all scattering images acquired from 616 Red Delicious apples. A firmness prediction model was developed by multi-linear regression against the Lorentzian parameters. Promising firmness predictions were obtained with the correlation coefficient of 0.82 and the standard error for validation of 6.39 N.