基于图像自动识别技术的雨滴谱测量方法

摘要

As the photograph-based raindrop size measurement suffers from various manual interactions,this paper proposed a new method for measuring raindrop size distribution based on computer vision.This approach included two steps:firstly,it ap-plied the anisotropic filtering to reduce the image noise and recognized raindrops automatically by incorporating the techniques such as segmentation and morphologic operations.Based on the recognition results and the calibrated 2D-3D transfer equation, this method computed raindrop size distribution directly from all raindrops.Experimental results demonstrate that the proposed method achieves a good performance on detecting raindrops and is capable of measuring raindrop size distribution with wide bandwidth.%针对摄像方式的雨滴谱测量方法中存在人工参与多的缺点，提出了一种基于计算机视觉的雨滴谱测量方法。该方法分为雨滴目标自动识别及雨滴谱计算两个步骤。采用各项异性扩散滤波的方式对雨滴图像进行降噪，结合图像分割与形态学等操作达到识别个体雨滴目标的目的。基于识别结果以及雨滴平面—立体转换方程，进一步计算得到实时雨滴谱。实验证明，该方法对雨滴群中单独个体的检测较为准确，并能够测量具有较广谱宽的雨滴谱分布。