The lack of mature detection methods for the high-precision measurement of chip warpage during thermal cycling affects the optimization of chip performance and improvement of packaging processes. To address this issue, we design and develop a three-dimensional (3D) detection system based on monocular fringe structured light, combined with infrared temperature detection. This study focuses on 3D reconstruction and warpage calculation methods for chip surfaces under varying thermal environments. By integrating fringe Gray code and the phase-shifting method, high-precision 3D reconstruction of point cloud data on chip surfaces is achieved. Additionally, the Robust Gaussian Lowess fitting algorithm is applied to fit the surface of point cloud data, reducing the fluctuation caused by noise points by approximately 30%, enabling accurate warpage calculation for chips. Experimental results demonstrate that the proposed method effectively suppresses noise and outlier effects in point cloud data, significantly improving the robustness and accuracy of surface fitting. It shows particularly high stability under thermal cycling conditions. The optical measurement system designed in this study is validated to meet micron-level warpage measurement requirements, providing reliable technical support for chip design and packaging process optimization. The system exhibits excellent application potential in complex thermal cycling test scenarios and offers a new solution for the high-precision quality detection and control of electronic components.