An image processing method comprises: acquiring an actual image of a specified target from a video stream collected by a camera; identifying an area not shielded by the VR HMD and an area shielded by the VR HMD of the face of the specified target from the actual image, and acquiring first facial image data corresponding to the area not shielded; obtaining second facial image data matching the first facial image data according to the first facial image data and a preset facial expression model, wherein the second facial image data correspond to the area shielded; and fusing the first facial image data and the second facial image data to generate a composite image. An image processing device comprises a first acquiring unit, an identifying unit, a second acquiring unit and a generating unit, and is for performing the steps of the method described above.