The Research Of Illumination Consistency In Augmented Reality

In order to realize the seamless synthesis of virtual and reality in the system of Augmented Reality, the problem of illumination consistency must be resolved. The illumination consistency refers to the consistency of illumination of the virtual objects and the real scene. The virtual and the real objects have the same effects of shade and shadow, and the augmented effect is enhanced, so users can not feel the inconsistency of the virtual and real. The key issue of illumination consistency is problem of acquisition of illumination information in the real scene, and the realistic augmented reality system could be generated by accurate illumination information.This paper is based on analysis of the current illumination consistency research in Augmented Reality, and proposed one method that combination of image-based lighting technique and high dynamic range image to resolve the problem of illumination consistency. Two methods are proposed to generate the high dynamic range image.1. High dynamic range images generated relying on multiple exposure imagesThe illumination information real-time change with the real scene in the Augmented Reality system, high dynamic range images need to real-time update. This paper proposed a method of use multiple low dynamic range images of different exposures synthesis for high dynamic range images. The method is based on albedo of virtual object and the function of camera imaging calculates the dynamic range that each frame required. Then this paper use bracketing to set up exposure time and generate high dynamic range images of dynamic scene which the system is needed. The light source information is obtained and radiance map generate from the High Dynamic Range image, the virtual object with the illumination effects is displayed by using tone mapping technology.2. High dynamic range images resumed relying on inverse tone mapping and light source imagesIn order to realize real-time and universal, this paper proposed a method of the low dynamic range image expanded and generated the high dynamic range image. The method : first, the preliminary high dynamic range image generated from low dynamic range image which calculated by inverse tone mapping, and then in order to compensate the inadequate of high light regions which caused by inverse tone mapping, the light source map is generated by acquiring light source regions and calculating their intensity to resolve this problem, last, high dynamic range image is correctly generated by interpolate light source map and preliminary high dynamic range image. For better real-time and stability of the system, we use the frame difference to determine whether the need to update the illumination information.The experiments are respectively executed from the two methods, the experiments demonstrate that the proposed two methods can be more real-time, realistic realize the virtual and real illumination consistency. The second algorithm is not subject to hardware constraints and it is universal and practicality.