空间频域成像在农产品品质检测中的应用现状与展望

空间频域成像，作为一种新兴的光学成像技术，具备宽场非接触、成像深度辨析和有效信号增强等特点，能够提供与组织物理结构、化学成分相关的信息，被广泛应用于农产品组织光学特性表征和品质无损检测等领域。该文首先概述了空间频域成像技术的起源和发展，继而阐明了该技术的工作原理，包括光在生物组织中的传输理论与正向问题、测量与数据处理、逆向反演，然后描述了该技术的多种实施方式，如常规空间频域成像、多光谱空间频域成像、高光谱空间频域成像以及高频空间频域成像，并总结其在苹果、梨、桃等农产品组织光学特性表征和品质检测方面的应用现状，最后讨论了该技术面临的挑战，如测量双层/多层农产品组织光学特性时误差较大、测量深度局限于毫米级、缺乏标准化的光学参考样本、检测耗时较长等，以期为该技术在未来的研究提供参考。

Application status and perspective of spatial-frequency domain imaging in quality evaluation of agricultural products

Light interaction with plant tissue varies significantly in different components with the structural, chemical, and optical characteristics in most agro products at the microscale. In light-tissue interaction, the tissues can generally be treated as being primarily composed of absorption and scattering particles, and thus the light propagation through tissues can be simplified as mainly involving the process of photon interactions with the absorption and scattering particles. When entering the tissue, the light can be absorbed and/or scattered, represented by the absorption coefficient (μa) and reduced scattering coefficient (μ′s), respectively. Quantification of optical properties (i.e., μa and μ′s) can greatly contribute to clarifying the measured data, optimizing optical devices, and finally improving the quality and safety assessment of agro-products. Alternatively, Spatial-Frequency Domain Imaging (SFDI) has widely been used to measure the optical properties, and then to evaluate the quality/safety of agro products last decades, showing the wide-field and noncontact imaging, depth- and resolution-varying, as well as signal enhancement. SFDI can also be used to reconstruct the three-dimensional distribution of optical features related to tissue physicochemical properties in the field of nondestructive detection. This study first overviewed the origins and development of SFDI in the field of agricultural engineering, and then introduced the main working principles of SFDI, including system components, light propagation model, data measurement and processing, and inverse algorithms for optical property estimation. Specifically, the SFDI was first applied to the nondestructive detection of bruising on Golden Delicious apples in 2007, indicating a better performance to distinguish the bruised apple from the sound one. The SFDI system is mainly composed of a light source, a digital projector, a CCD camera, a wavelength selective device, and a sample stage, the former three of which are the core components to directly determine the quality of structured illumination and collected images, as well as the testing efficiency. Calibration is also required for the SFDI system with the standard samples before evaluation. The specific procedure is followed. The images of target samples are first captured by the SFDI system. The light uniformity correction, image demodulation, system response calibration, and surface profile correction are then conducted to obtain the diffuse reflectance images for the quality and safety evaluation directly, or for the optical property estimation coupled with inverse algorithms. After that, the application status of SFDI was summarized in the field of agricultural engineering, including the measurement of optical property and quality/safety assessment of several thin-skinned fruits, such as apple, pear, kiwifruit, cucumber, and peach. The challenges and future perspectives of the SFDI technique were also presented eventually. Nevertheless, the current SFDI technique is derived mostly from the diffusion approximation, thereby hindering the application easy to introduce large measurement errors. There are great challenges when measuring the optical property of two- and multi-layered agro products. It is also lacking a standardized optical system for accurate estimation of the optical property. The SFDI presents better performance in the depth-varying detection, but the penetration depth is a bit shallow limited to the millimeter level. Moreover, the demand for portable handheld devices of the SFDI technique is ever increasing in recent years. This review can provide a critical overview of the development of the SFDI technique for better understanding in the field of agricultural engineering.