Elbow-shaped workpiece is widely used in chemical and metallurgical production. Its cross section parameter requires extremely high accuracy but the workpiece has irregular internal and external shapes which result low manual measurement efficiency, poor detail capture and high error rate which is difficult to realize the requirement of industrial online real-time detection. Therefore line structured light scanning to obtain three-dimensional point cloud data of the shape of the elbow pipe is used to measure the section of the elbow pipe in real time. Calibrated linear structured light three-dimensional measurement platform scan the elbow pipe model moving at about 4.3cm/s and obtain its point cloud data. The point cloud data registration closest point iterative method, Gaussian noise removal, improved simplified angular deviation method then using the Delaunay triangulation algorithm to reconstruct the point cloud in three dimensions and measuring the width radius and inner diameter of the reconstructed model. It takes about 20s and the measurement error is less than 0.05mm. The method has high precision and accuracy which greatly improve the efficiency of workpiece detection. It has great industrial application prospect.