The grinding wheel, composed of abrasive grains, bonding agents, and pores, plays a crucial role in grinding processes, for the grain distribution, size, and orientation impact the grinding quality directly. Accurate surface modeling is essential but hampered by the absence of unified standards and challenges in measuring surface characteristics. Therefore, this paper presents a novel method for measuring grinding wheel surfaces using image recognition and clustering algorithms. These techniques effectively identify abrasive grains in microscopic images and extract key features including distribution density and protrusion height to form the abrasive grain database. The validation based on real microscopic images demonstrated that, compared with traditional empirical models, the proposed model shows higher efficiency and accuracy in surface characterization. The integration of these algorithms significantly enhances the accuracy of surface modeling.