Development And Application Of A Handheld Near-infrared Spectrometer Based On A Linear Variable Filter For Measuring The Internal Quality Of Fruit

Consumers are becoming increasingly demanding with regard to fruit quality. Near infrared spectral analysis offers the advantages of being noninvasive and efficient. It has long been used to evaluate internal quality of fruit and has been proved to be the most appropriate technique for this application. Fruit may be assessed using near-infrared spectroscopy in the orchard, in the storehouse, on the grading line or in the market. Therefore, most applications are on-site measurements, which require portable tools. The linear variable filter can replace a grating as the light-dispersion component. A linear variable filter was fabricated specifically to work in the 620–1080 nm region. The purpose of this research is to develop a handheld spectrometer with a linear variable filter to predict the internal quality of fruit. The major work and innovations are introduced as follows:1) Analysis of impact of instrument parameters(incident light, distance between filter and detector, reflectivity of the detector) on spectral performances of the linear variable filter. Performances of a linear variable filter are strongly dependents to instrument parameters. Impact of Incidence light cone on spectral performances of the linear variable filter was analyzed with theoretical analysis and ZEMAX optical models. Evolution of the spectral performances with the distance between the filter and the detector was also analyzed. Impact of the reflectivity of the detector was also analyzed.2) The design of the opto-mechanical system. Using an optimised collimator, a compact optical engine was obtained, providing a high degree of stability and a resolution of 1.5% center wavelength. The measuring head was designed to use interactance mode for acquisition of spectral information. The instrument is compact and ruggedness.3) The design of the electronic system. Using a CMOS detector, the electronic system had a low noise and low power consumption. The designs of the driving system, the data acquisition system, the power supply system are detailed. The spectrometer had a low power consumption and performed with a signal-to-noise ratio of up to 5000.4) Application of the spectrometer to predict the internal quality of fruit. The sugar content in pears and apples was predicted with the spectrometer prototype. Method of selecting calibration samples、outlier sample eliminating methods and four pre-treatment methods were used to optimize prediction models. Promising prediction results were obtained. These results demonstrated that this spectrometer is highly suitable for predicting fruit internal quality.