Radiometric Compensation for Ubiquitous Projection

Ubiquitous projection, meaning being able to project an image anywhere, is no longer a fiction due to the miniaturization of projectors. However, flat surfaces in our living environment to be used as the replacement of a projection screen for pico or handheld projectors are not necessarily white. To improve the image quality for ubiquitous projection, a perceptual radiometric compensation system is presented in this dissertation to counteract the effect of color projection surfaces on image appearance. In the first part of the dissertation, a perceptual radiometric compensation method is presented to counteract the effect of color projection surfaces on image appearance. It reduces color clipping while preserving the hue and brightness of images based on the anchoring property of human visual system. In addition, it considers the effect of chromatic adaptation on perceptual image quality and fixes the color distortion caused by non-white projection surfaces by properly shifting the color of the image pixels toward the complementary color of the projection surface. User ratings show that the method outperforms existing methods in 974 out of 1020 subjective tests. In the second part of the dissertation, a visibility enhancement method is presented to counteract the visibility loss effect caused by the brightness scaling in the process of radiometric compensation. Based on JND theory and the HVS response model, the method effectively enhances the visibility of image details in dark regions without affecting the perceptual contrast of bright regions. The method also applies appropriate counter shading to eliminate halo effect and, meanwhile, enhance perceptual contrast of the brightness-scaled image. Experimental results are provided to show the performance of the method. In the last part of the dissertation, a simplified CIECAM02 model is presented to reduce the computational burden caused by the nonlinear color transforms of CIECAM02. Experimental results show that the simplified model reduces the computation time by 50% with only 2.3% approximation error. The speedup is important for practical applications.