In order to realize the full range and high-precision 3D measurement of small objects, a high-precision 3D measurement system using structured light illumination is constructed. The double telecentric lens, phase solving method and multi-projection point cloud fusion algorithm used in the system are studied. Firstly, the image data is collected by dual telecentric measuring head, then the phase is solved by multi-frequency phase shift and complementary gray code phase shift, and the performance of the two methods in terms of reconstruction accuracy and reconstruction efficiency is analyzed and compared. Finally, the phase filtering method, the optimized statistical filtering method and the multi-projection point cloud fusion matching correction method are proposed for specific point cloud noise. The experimental results show that the system should choose different phase solving methods according to different scenarios. Compared with the single projection binocular system, the multi-projection based system can obtain the full picture of the target, and the precision of the plane and height measurement is less than 10 m. After GPU acceleration, the measurement speed is increased by 218 times. The system can basically meet the requirements of high precision and high efficiency of high precision industrial measurement.