摘要
双目立体影像以平面作为零视差表面,采用两台透视相机生成左右眼图像。展览展示应用通常将上述方法生成的立体影像沿长宽两个方向均匀映射到柱面屏幕,使观察者对立体影像的深度感知产生变形。描述柱面条件下立体影像生成、显示和感知过程的几何分析模型,利用该模型对均匀映射下柱面中心的立体影像深度感知进行分析。理论模型的分析结果表明,均匀映射方法会造成柱面中心深度感知的非线性放大,引起深度平面弯曲等现象,影响观察者对场景的认知。数据实验结果表明,均匀映射条件下经验法则限定的深度范围仍旧适用;根据模型计算的安全深度区间感知映射结果,内容生成者可进一步减小成像时的最小深度边界,获得更加突出屏幕的立体视觉效果。
Stereoscopic images are commonly captured by two perspective cameras with one plane as imaging surface. These stereoscopic images are presented on cylindrical screen through simple uniform mapping, which introduces distortion of depth perception. In this paper, we show a geometry model formulating the process of stereoscopic image production, presentation and perception on cylindrical screen. Using the model, we analyze the depth perception with uniform mapping at center of cylinder screen. The analysis results illustrate perceptional depth non-linearity amplification and perceptional depth plane curvature, which leads false perception for the scene. The experiment results support that the well-established rules of comfort zone around the screen still works using uniform mapping. Furthermore, exploiting the model' s depth interval mapping, stereoscopic image designer can relax the lower bound of object depth in production procedure for achieving strong visual effects.
出处
《计算机应用与软件》
CSCD
北大核心
2013年第1期16-20,33,共6页
Computer Applications and Software
基金
国家自然科学基金项目(60803064)
关键词
立体成像
深度感知
计算机视觉
Stereo production Depth perception Computer Vision