Illumination Estimation Based On Image Decomposition

Augmented reality technology can superimpose computer-generated virtual objects into real scenes.With the development of computer graphics technology in recent years,its applicability has been significantly enhanced.The key to realizing real enhanced display effects is that virtual objects need to show shadows and colors under real lighting.The lighting information can be estimated more accurately by using different kinds of reflected light information in the scene.First of all,based on the pixel clustering image decomposition method,the specular reflection of the light in the image can be decomposed,and the result is optimized by using high-emphasis filter.After the specular reflection image and the diffuse reflection image are obtained,the intrinsic image decomposition is used to decompose the diffuse reflection image.It is further decomposed into an albedo map and a shadow map.Furthermore,a Markov random field model is applied to the image,and the depth of pixels in the image is estimated using the three-dimensional positional relationship between superpixels.Finally,a global rendering model was used to develop an optimized rendering-based model to estimate the lighting information.Combining the obtained depth map and the decomposed image results,the light source position and the color of the light source of the scene can be estimated.After obtaining the light source information,the virtual object can be relighting in the scene.Using the above method to perform virtual object rendering experiments,the results show that the method proposed in this paper can effectively realize the illumination estimation and virtual object rendering in Reflectence surface scenes,and can obtain more realistic rendering results.