针对二维多视点图像阵列由于采集设备限制而难于获取的问题,采用单个相机结合摄影轨道的方法进行采集。为了提高采集效率和满足不同显示设备对图像源的需求,提出一种基于深度图绘制(DIBR)技术的二维虚拟视点合成算法,在水平、垂直以及斜线方向实现了相邻参考图像间的虚拟视点合成。实验结果给出了实际采集到的二维多视点图像阵列以及合成的虚拟视点图像,并分别从图像质量和立体感两个方面对虚拟视点图像进行评价,评价结果表明,虚拟合成后的图像具有连续的视差变化可以真实再现拍摄对象的结构信息。
In an effort to effectively solve the hard pickup problem of the two dimensional multiview image arrays due to the pickup device limitations, we used a single camera and a photography track to finish the capture process. To improve the pickup efficiency and meet the demands of different display devices for the image, a synthesis algo?rithm based on the Depth Image Based Rendering (DIBR) technology was proposed. The purpose is to obtain the two dimensional virtual viewpoint images through the adjacent reference images in the horizontal, vertical, and di?agonal directions. The captured two-dimensional multiview image array and the synthesized virtual viewpoint images were present in the experimental results;an evaluation of the synthesized virtual viewpoint images from the two as?pects of image quality and stereo sense were performed. The results show that the synthesized virtual viewpoint ima?ges have continuous parallax changes and can truly represent the structure information of the subjects.