Abstract: Abstract: Determination of internal quality of fruit and vegetable with a suitable technique is crucial for processing detection and quality control. While substantial progress has recently been made in the miniaturization of near-infrared (NIR) spectrometers, there remains continued interest from end-users and product developers in pushing the technology envelope toward even smaller and lower cost analyzers. The potential of these instruments to revolutionize on-site applications can be realized only if the reduction in size does not compromise performance of the spectrometer beyond the practical need of a given application. In this paper, the working principle of a novel, extremely miniaturized NIR spectrometer is presented. The ultra-compact spectrometer relies on digital micromirror device (DMD) technology for the light dispersing element. DMD is a two-dimensional array of electro-mechanical mirror elements whose surface normal angles can be controlled. Digitally programmable DMD can set the spectral resolution and wavelength range according to user needs, adjust the integration time, and adapt the luminous flux. The system design with DMD and single-pixel InGaAs detector can significantly reduce the cost, and meanwhile ensure the detect precision. The DLP (digital light procession) NIRscan module is used as spectrometer optical engine in the miniaturized system. In the specific implementation, a sample is placed against the sapphire front window of the reflectance head. During a scan, the sample absorbs a specific amount of NIR light and diffusely reflects the non-absorbed light into the system. The illuminating lamps are designated as lens-end lamps because the front end of the glass bulb is formed into a lens that directs more lights from the filament to the sample test region. The collection lens gathers collimated light from a 2.5 mm diameter region at the sample window. The handheld system supports the following modes of operation: USB (Universal Serial Bus) connection and Bluetooth for 2 communication channels. Special analyzer software was developed for quality inspection based on multithread programming technology. The advantages of this software are presented by the process of modular design, including software system initialization, information communication, information interaction, spectral data acquisition and processing, spectral curve real-time display, quality index calculation, and statistics and save of detection results. Miniaturized handheld NIR spectrometer was developed and used to acquire reflectance spectra from fruit and vegetable samples in the wavelength range of 900-1 700 nm. In order to verify the design and performance, tomato was selected as research object. A total of 78 tomato samples were randomly divided into 2 subsets. The first subset was called the calibration set with 52 samples and used for building model, while the other one was called the prediction set with 26 samples and used for testing the robustness of the model. In the process of model establishment, a simplified strategy was proposed. Firstly, characteristic spectrum bands were selected to remove the uninformative variable and the low-correlation band. And then feature wavelengths were optimized to eliminate the collinearity relationship in the spectral data. Finally, simplified model was built, which had good robustness and stability. Synergy internal partial least square (siPLS) and successive projections algorithm (SPA) were sequentially applied to calibrate models. The siPLS was applied to select an optimized spectral interval and an optimized combination of spectral regions selected from informative regions in model calibration. The subsequent application of SPA to this reduced domain could lead to an efficient and refined model. The measurement results of the final model were achieved as follows: correlation coefficient (Rp) was 0.899 and root mean square error of prediction (RMSEP) was 0.133% for soluble solid content in tomato, and Rp was 0.886 and RMSEP was 2.508 mg/kg for lycopene content in tomato. The results will provide the method reference for rapid, non-destructive, and on-site detection technology and equipment of fruit internal quality.