The accurate inspection of surface flatness is important for the quality inspection of ceramic tiles. According to quality standards, the flatness value is evaluated based on the surface information of ceramic tiles. Current industry practice relies on manual inspection or displacement sensors to collect surface information, which has limitations in accuracy due to incomplete surface information. Structured light 3D scanners can offer more comprehensive surface information and more accurate flatness inspection. However, processing large-scale point cloud data from the 3D scanner in real time presents a significant challenge. To address this challenge, we propose a real-time flatness inspection algorithm for ceramic tiles using the structured light 3D scanner (FIT3D). FIT3D utilizes a support vector machine (SVM) model to deal with the large-scale point cloud and determine the minimum zone planes. The convex hull theory is then combined within the model to speed up the algorithm. The effectiveness of the algorithm is demonstrated through studies on several datasets in literature and a real-world case study.