A 3D reconstruction method is provided. Positioning signals are received by a signal receiver at a first and a second time spots to determine a HMD displacement vector and a HMD rotation amount. A first and a second images are retrieved by a first camera to determine a first camera rotation amount. A relative rotation amount and a relative displacement vector between the HMD and the first camera are calculated. A first camera displacement vector of the first camera is calculated according to the HMD displacement vector, the HMD rotation amount, the relative rotation amount and the relative displacement vector. Depth information of the first and the second image is obtained based on the first camera displacement vector and the first camera rotation amount. 3D reconstruction is performed according to images retrieved by the first camera and the depth information.