The morphology of particles is usually detected by microscopic imaging methods such as scanning electron microscope (SEM). However, this type of direct observation relies on expensive equipment and requires complex preparations. Inspired by the transit method in astronomy, a characterization system based on the scattering patterns change due to the relative motion between the particle and facula was proposed in this work. The scattering patterns of particles could be recognized and analyzed automatically to detect the particle shape. The flow trajectory was studied by the optical fluid field couple (OFFC) model proposed in this work to calculate the size and specific surface area of the particle. For getting more accurate results, particle surface roughness was measured by image recognition to correct the results of specific surface areas. The accuracy of the classification machine learning (CML) model was about 97%, and the errors of the specific surface area results were lower than 10%. Comparing with the previous work, the proportions of different particles in mix sample could be detected successfully by this system with a maximum error less than 6%. This work can offer a valuable reference for the fields of characterization morphology systems based on scattering.

颗粒的形貌通常通过扫描电子显微镜（SEM）等显微成像方法检测。然而，这种直接观察依赖于昂贵的设备并且需要复杂的准备工作。受天文学中凌日法的启发，本文提出了一种基于粒子和光斑之间相对运动引起的散射模式变化的表征系统。可以自动识别和分析粒子的散射模式以检测粒子形状。通过本工作中提出的光学流体场耦合（OFFC）模型研究流动轨迹，以计算颗粒的尺寸和比表面积。为了获得更准确的结果，通过图像识别测量颗粒表面粗糙度，以修正比表面积的结果。分类机器学习（CML）模型的准确率约为97%，比表面积结果的误差低于10%。与之前的工作相比，该系统可以成功检测混合样品中不同颗粒的比例，最大误差小于6%。该工作可为基于散射的表征形态学系统领域提供有价值的参考。