适于多种反射现象的光照估计方法
|
摘要
增强现实技术是将虚拟物体叠加到真实环境中,为了使虚拟物体更逼真地融入到环境,就需要估计场景中的真实光照。文中针对同时存在漫反射和镜面反射的场景或只存在漫反射的场景,提出了一种适于多种反射现象的光照估计方法。该方法使用相机拍摄多幅视角图像,通过检测每幅视角图像的镜面高光点来识别场景中是否存在镜面反射。若不存在镜面反射,则采用单张漫反射图像估计光照算法进行多幅视角图像光照估计。反之,若存在镜面反射则以镜面反射提供的信息为先验知识,再结合漫反射光照估计方法得到最终的光照估计结果。实验结果表明,文中方法提高了多种反射现象的光照估计的准确性。
In an augmented reality system, the virtual objects are superimposed on the real scene. In order to integrate the virtual objects into the environment more realistically, it is necessary to estimate the illumination in the scene. An illumination estimation method was proposed, which was suitable for multiple reflection phenomena for scenes with only diffuse reflection or scenes with both diffuse and specular reflections. In this method, a camera was used to shoot multiple-view images, and the presence of specular reflection in the scene was identified by detecting the specular highlights of each view image.If there was no specular reflection, then a single diffuse image estimation lighting algorithm was used to perform multi-view images illumination estimation. If there existed specular reflection, the information provided by the specular reflection was a priori knowledge and then the diffuse reflection estimation illumination method was used to obtain the final result. The experimental results show that this method improves the accuracy of the illumination estimation of multiple reflection phenomena.
In an augmented reality system, the virtual objects are superimposed on the real scene. In order to integrate the virtual objects into the environment more realistically, it is necessary to estimate the illumination in the scene. An illumination estimation method was proposed, which was suitable for multiple reflection phenomena for scenes with only diffuse reflection or scenes with both diffuse and specular reflections. In this method, a camera was used to shoot multiple-view images, and the presence of specular reflection in the scene was identified by detecting the specular highlights of each view image.If there was no specular reflection, then a single diffuse image estimation lighting algorithm was used to perform multi-view images illumination estimation. If there existed specular reflection, the information provided by the specular reflection was a priori knowledge and then the diffuse reflection estimation illumination method was used to obtain the final result. The experimental results show that this method improves the accuracy of the illumination estimation of multiple reflection phenomena.
引文
[1] Jacobs K, Nahmias J D, Angus C, et al. Automatic generation of consistent shadows for augmented reality[C]//Conference on Graphics Interface, 2005:113-120.
[2] Akila Pemasiri, Chameera Wijebandara, Samee-ra Wijayarathna,et al. An online lighting model estimation using neural networks for augmented reality in hand-held devices[C]//International Conference on Advances in ICT for Emerging Regions, 2015:4-8.
[3] Natalia Neverova, Damien Muselet, Alain Tr′emeau. Lighting estimation in indoor environments from low-quality images[C]//Proc of ECCV, 2012:380-389.
[4] Yoo J D, Lee K H. Real time light source estimation using a fish-eye lens with ND filters[C]//International Symposium on Ubiquitous Virtual Reality, 2008:41-42.
[5] Chen Xiaowu, Wang Ke, Jin Xin. Single image based illumination estimation for lighting virtual object in real scene[C]//International Conference on Computer-aided Design, 2011:450-455.
[6] Liu Bin, Xu Kun, Ralph R Martin. Static scene illumination estimation from videos with applications[J]. Journal of Computer Science and Technology, 2017, 32(3):430-442.
[7] Qing Tian, James J Clark. Real-time specul-arity detection using unnormalized wiener entropy[C]//International Conference on Computer and Robot Vision, 2013:356-363.
[8] Salma Jiddi, Philippe Robert, Eric Marchand. Reflectance and illumination estimation for realistic augmentations of real scenes[C]//IEEE International Symposium on Mixed and Augmented Reality, 2016:244-249.
[2] Akila Pemasiri, Chameera Wijebandara, Samee-ra Wijayarathna,et al. An online lighting model estimation using neural networks for augmented reality in hand-held devices[C]//International Conference on Advances in ICT for Emerging Regions, 2015:4-8.
[3] Natalia Neverova, Damien Muselet, Alain Tr′emeau. Lighting estimation in indoor environments from low-quality images[C]//Proc of ECCV, 2012:380-389.
[4] Yoo J D, Lee K H. Real time light source estimation using a fish-eye lens with ND filters[C]//International Symposium on Ubiquitous Virtual Reality, 2008:41-42.
[5] Chen Xiaowu, Wang Ke, Jin Xin. Single image based illumination estimation for lighting virtual object in real scene[C]//International Conference on Computer-aided Design, 2011:450-455.
[6] Liu Bin, Xu Kun, Ralph R Martin. Static scene illumination estimation from videos with applications[J]. Journal of Computer Science and Technology, 2017, 32(3):430-442.
[7] Qing Tian, James J Clark. Real-time specul-arity detection using unnormalized wiener entropy[C]//International Conference on Computer and Robot Vision, 2013:356-363.
[8] Salma Jiddi, Philippe Robert, Eric Marchand. Reflectance and illumination estimation for realistic augmentations of real scenes[C]//IEEE International Symposium on Mixed and Augmented Reality, 2016:244-249.