Abstract: Raindrop spectrum is a critical parameter for describing the microphysical characteristics of precipitation, so its accurate measurement plays a pivotal role in enhancing radar-based quantitative precipitation estimation, investigating microphysical properties of precipitation, and understanding the evolution process of precipitation. To achieve synchronous measurement of raindrop size and terminal velocity, we propose an optical raindrop measurement method based on machine vision principles, design a system with a dual telecentric lens configuration, a telecentric light source, and a linear array camera as the core, and constuct a prototype of an optical raindrop spectrometer in this paper. The data processing and analysis system of the spectrometer, is based on the Microsoft Foundation Classes (MFC) framework, and a real-time automated raindrop measurement software is developed utilizing the camera software development kit (SDK) in conjunction with Halcon software. The Canny algorithm is employed for subpixel edge detection of the acquired images, interpolation algorithms are applied to restore raindrop contours, and finally based on the restored raindrop images, the size and terminal velocity of raindrops are computed, thus achieving automated measurement of raindrops. Calibration tests are conducted using steel spheres and water droplets, with calibration of the instrument's measurement range being performed using a standard gauge block. The calibration results show that the raindrop spectrometer has a measurement error of less than 21 μm and a velocity inversion error below 4.5% for steel sphere diameters between 0.6 and 5.0 mm, and the standard deviation for droplet measurements is 26 μm, indicating that the instrument can concurrently measure particle size and terminal velocity, and has good accuracy and consistency in measuring moving particles.

Key words: raindrop spectrometer, machine vision, edge detection