Efficient stereo rendering of large 3D datasets based on binocular suppression

As the demand for stereo 3D content increases in the various fields, many types of devices for displaying stereo contents have been developed such as TVs, laptops, and mobile devices. Thus, efficient rendering methods that can process massive stereo data on those devices are becoming more important as well as reducing a feeling of visual discomfort of the viewer. In this paper, we analyze binocular suppression in an aspect of human visual perception, and present a point-based stereo rendering method to process high-resolution 3D data based on binocular suppression. To reduce visual discomfort which occurs in the process of the stereo rendering, we develop a simplification method by extending spectral sampling. Using the adaptive sampling, we can improve the rendering quality, especially in the visually sensitive areas as well as reducing the size of datasets.