GPU-Based Adaptive Per-pixel Photometric Calibration for Tiled Multi-projector Autostereoscopic Displays
There are mainly two kinds of traditional photometric calibration methods for tiled multi-projector autostereoscopic displays. One is the LAM (Luminance Attenuation Map) method which needs high cost to calculate the intensity transfer function (ITF), and the other is the edge-blending method which addresses only the color difference in the overlap regions. In order to eliminate the drawbacks of the aforementioned methods, a GPU-based adaptive perpixel photometric calibration approach is presented in this paper. The proposed approach, at first, generates the reference masks for the reference colors by several iterations of a feedback process. For each image, a photometric calibration mask is computed from the reference masks, and each value of it is related with the luminance of the corresponding pixel in the geometrically corrected image. At last, the photometric calibration mask for each image acts as the pixelwise scale factors for scaling the luminance response in order to achieve the photometric uniformity. The experimental results show that the GPL-based adaptive per-pixel photometric calibration approach can achieve visually seamless displays in tiled multi-projector autostereoscopic display systems.
autostereoscopic display feedback geometric calibration GPU photometric calibration tiled display
Yanxia Zhou Kaihuai Qin Jianli Luo
Department of computer science and technology Tsinghua University Beijing, China Military operations Department of computer science and technology Tsinghua University Beijing, China
国际会议
2010 International Conference on Future Information Technology(2010年未来信息技术国际会议 ICFIT 2010)
长沙
英文
375-379
2010-12-14（万方平台首次上网日期，不代表论文的发表时间）