Abstract: To achieve automatic mesh size detection for vibrating screening equipment with high precision, an in-situ mesh size measurement system was designed, using image acquisition devices and image processing algorism based on parallel light visual scanning. The overall dimensions of the image acquisition device composed of a parallel light module and a visual scanning module were 450 mm × 340 mm × 106 mm. The device can capture sieve mesh images by illuminating the sieve mesh with parallel light and a contact image sensor. A feeding scanning mechanism was used to expand the field of view and improve image resolution. During image processing, an adaptive sieve mesh contour extraction algorithm was developed based on an adaptive threshold binarization method which combined global and local thresholds for precise binarization within local windows, and an adaptive sub-pixel edge interpolation positioning method which dynamically adjusted the number of interpolation points for the edges of sieve meshes with different sizes to enhance noise removal and positioning accuracy. The system's performance was validated through accuracy evaluation and in-situ sieve detection experiments. The results showed that: 1) The system offered a single detection area of 320 mm × 320 mm, a resolution of 0.010 6 mm, and a single scan time of 30 s; 2) Using sieve mesh standard samples, indexes measured by the system such as mean mesh size limit deviation were superior to the requirements specified in the existing standard, meeting the performance requirements for the mesh size detection instruments. 3) This system demonstrated good applicability and stability, and met the in-situ detection requirements of the mesh size of tobacco vibration screening equipment. 4) When installing the image acquisition device, a slight tilt (tilt angle < 1.5°) between the parallel light source module and the visual scanning module or a distance (distance≤20 mm) between the two modules had a negligible impact on the qualification of the detection results. This technology provides support for the enhancing sieve mesh detection efficiency and the accuracy of physical tobacco indexes such as tobacco strip size and cut tobacco structure.