Self-Adaptive Radiometric Compensation For Dynamic Environments

Augmented Reality develops from Virtual Reality.It increases the user's perception of the real world using information provided by computer system.It combines computergenerated virtual objects,system message with the real scene to achieve the enhancement of reality.And augmented reality,compared with virtual reality,has the advantage of more realistic and more interactive.It has been widely used in engineering,medical, military,education,entertainment and other fields.The traditional display technology for augmented reality is Head-Mounted-Display (HMD),which is heavy and inconvenient.Projection display as a new means of display, with the advantage of convenient,fast and clear,has been widely used in daily lives. And it can be used in augmented reality.However,the quality of projection display is affected by ambient light and projection surface.For mobile projection equipment,the great impact is the need of white projection surface.Researchers proposed radiometric compensation technology to make it possible for projection on non-white surface with the usage of projector-camera system.But it introduces other problems.First,there's a long calibration process before projection works;second,re-calibration which will interrupt the projection process,is needed once the environment changes.These makes the projection display complicated,and limits radiometric compensation's widely usage.In this paper,we first study the geometric mapping for projector camera system. We create the geometric mapping by projection chessboard pattern.Then we propose a radiometric compensation method for static environment,which is simple,easily configured, and no special hardware requirements.However,this approach can't work in dynamic environment.By studying the photometric model of projector-camera system, we found that the color mixing matrix of the system reflects the physical attributes of projector and camera,and is independent of environment.Also we found that there is one global color-mixing matrix for the system.Based on the above observation,we propose an improved system model,which takes ambient light,reflectivity of projection surface, and the system vignetting effect into account.This model is more accurate and use of less resources.We implement a radiometric compensation for dynamic environments based on the improved system model.We use camera to monitor the environment change,and recalibrate the system to update the internal parameters.Experimental result shows that the improved model reduces the use of resources and speed up the process of calibration.For dynamic environment,we project just one image to finish the recalibration process,which is fast enough for adaptive compensation.