【摘要】 柱透镜光栅的计算机辅助立体成像的基本原理是将二维序列图像进行切分和重组,使之以三维模型的形式展现,使合成的图像在柱透镜光栅下观察时具有"纵深感"。该项技术在广告、摄影、电影制作中具有广泛的应用。但是长期以来,在图像切分时并不能保证像素一定被整除,这意味着剩余像素的出现,合成的立体图像必然存在误差。如何处理剩余像素的问题一直是个难题,阻碍了该技术的推广。针对立体图像生成方法中的剩余像素问题,详尽地分析了误差的形成,并提出了强行取整算法和误差分散算法两种误差处理方法。实验表明,这两种算法能较好的处理剩余像素,增强了合成图像的立体效果,这将有助于立体成像技术的进一步推广。更多还原

【Abstract】 The facture principle of computer aided lenticular screen covering color stereo image is analyzed, the stereoscopic system presents the viewer an image of a 3D object such that it appears to have "real" depth. This technique has many applications such as 3D entertainment, but in a long time, it has limited use areas. The defect of residual pixel in image processing greatly affects the visual impression of the stereo image. According to the question, the forming reason of the error is investigated, and two new methods of error processing are proposed:forced integer method and error dispersal method . Experiments indicate that the error processing methods have gained satisfactory effect, it will contribute to the popularization of this technique.更多还原