Development of a multispectral spatial-frequency domain imaging system for property and quality assessment of fruits and vegetables

Abstract

Thanks to the unique features of wide-field, label-free and depth-resolved imaging, spatial-frequency domain imaging (SFDI) technique has witnessed rapid and considerable progress in biomedical, agricultural and food engineering. This paper reports on the design, characterization and calibration of a multipurpose, multispectral SFDI system (550, 600, 630, 675, 710 and 730 nm) for property and quality assessment of fruits and vegetables. The system was mainly composed of illumination unit, imaging unit and sample stage. Customized software was developed to synchronously control pattern generation and shift, light illuminating, image acquisition and saving. Multiple parameters, including field of view, spatial resolution, uniformity, linearity, repeatability and stability, were used to characterize or calibrate the multispectral SFDI system from a systematic view, based on a set of experiments using optical phantoms. The results indicate that the constructed multispectral SFDI system can be used to collect optical property parameters of common fruits and vegetables after calibration. Then the multispectral SFDI system was applied to measure the optical properties of six types of fruits and vegetables (apple, pear, cucumber, tomato, white and green radish), and the measured optical property mappings were utilized for detecting early-stage bruise, which cannot be observed by naked eyes or even conventional uniform lighting imaging. The results demonstrated that the multispectral SFDI system was capable of measuring optical properties, and the reduced scattering coefficient mapping was superior to the absorption coefficient mapping in detecting early-stage bruise. Selection of effective characteristic wavelength could further enhance the bruise detection.