Laser composite imaging 3D reconstruction is achieved through laser scanning and image processing algorithms to achieve high-precision 3D reconstruction of objects or scenes. However, in practical applications, due to the interference of lighting intensity and noise, technology is difficult to accurately capture pixel details, textures, and brightness features, resulting in a certain deviation between the reconstructed 3D image and reality. To this end, a three-dimensional reconstruction method for laser composite imaging under weighted Gaussian matching filtering is proposed. Based on the mechanism of laser composite imaging, a scanning equation is established to generate laser images. The image noise is smoothed using neighborhood averaging method, and pixel contrast is matched through kernel function and convolution operation. A three-dimensional reconstruction model is established using Markov random field. After distance fidelity and regularization operation, the three-dimensional reconstruction of laser composite imaging is achieved. The experimental results show that after weighted Gaussian matching filtering, the reconstructed results can realistically display the three-dimensional images of objects and scenes, with small errors and high structural similarity.