|Peng, Yankun - MICHIGAN ST UNIVERSITY|
Submitted to: Proceedings of SPIE
Publication Type: Proceedings
Publication Acceptance Date: October 4, 2004
Publication Date: December 1, 2004
Citation: Peng, Y., Lu, R. 2004. A liquid crystal tunable filter based multispectral imaging system for prediction of apple fruit firmness. Proceedings of SPIE. 5587:91-100. Interpretive Summary: Firmness is a key parameter for determining the overall quality of apples and many other fruits during harvest, postharvest handling and storage, processing and retailing operations. The Magness-Taylor firmness tester--the current industry standard technique--measures forces required for a mechanical probe to penetrate fruit flesh for a specified depth, which is destructive and, thus, not suitable for grading and sorting apples. This research reports on the development of a nondestructive optical technique, which is based on our proposed concept of measuring light scattering in apple fruit at selected wavelengths, for predicting fruit firmness. We developed a multispectral imaging system, which incorporated a low cost digital camera and a liquid crystal tunable filer that allows us to identify a set of optimal wavelengths over the near-infrared region. An improved mathematical model was proposed for describing light scattering profiles from apple fruit. A set of seven wavelengths were identified to be best for predicting the firmness of Red Delicious apples. The multispectral imaging system was able to predict fruit firmness with the correlation coefficient of 0.82 and the prediction error of 6.6 N, which was significantly better than near-infrared technique. The research provides critical information and a means for selecting optimal wavelengths for obtaining improved predictions of fruit firmness. Research demonstrated the technical feasibility of developing a low cost firmness sensing system, which will facilitate eventual transfer of the technology from laboratory to commercial applications. The technology is needed for assuring the quality and consistency of individual fresh fruit delivered to the marketplace, thus helping U.S. fruit growers become more profitable under the increasingly intense global competition.
Technical Abstract: Firmness of apple fruit is an important quality attribute, which varies greatly in the same lot of fruit due to such factors as climatic condition, cultural practice, harvest time or maturity level, and postharvest handling and storage. This research developed a compact multispectral imaging system with a low cost digital camera and a liquid crystal tunable filter (LCTF), and proposed a modified Lorentzian distribution (MLD) function to describe scattering profiles acquired from Red Delicious apples. The LCTF, which allows for the rapid, vibration-less selection of any wavelength in the visible/near-infrared range, was used to find optimal wavelengths over the spectral region between 650 nm and 1,000 nm for predicting apple fruit firmness. Radial scattering profiles were described accurately by the MLD function with four profile parameters for wavelengths between 650 nm and 1000 nm at an interval of 10 nm. Multi-linear regression (MLR) and cross-validation were performed on relating MLD parameters to fruit firmness. The prediction model gave good firmness predictions with the correlation coefficient (r) of 0.82 and the standard error of validation (SEV) of 6.64 N, which were considerably better than those obtained with visible/near-infrared spectroscopy.