Illumination estimation of augmented reality virtual human

Illumination consistency is an important aspect to implement virtual-real fusion in augmented reality system. An effective illumination estimation method is significant in these systems. In order to achieve illumination consistency for the virtual human in augmented reality, a novel method is proposed for illumination estimation, using the property that the virtual human has a relatively fixed target. This method gets the normal data through the three-dimensional reconstruction for the target, and overcomes the difficulties in the process of obtaining normal data in the general illumination direction estimating methods. In order to improve the computing efficiency of the system, the continuous images are replaced by discrete sampling points in analysis. Some experiments are designed to show that the proposed method for the virtual human in illumination direction estimation is feasible.