摘要
虚拟视点生成是3维视频会议等应用领域中的关键技术,为了快速高质量地进行任意视点绘制,提出了一种基于深度图像绘制(DIBR)和图像融合的新视点生成方法,该方法首先对参考图像进行预处理,包括深度图像的边缘滤波和参考图像规正,以减少目标图像中产生的较大空洞和虚假边缘;然后利用3维图像变换生成新视点图像,并用遮挡兼容算法对遮挡进行快速处理;接着再对两幅目标图像进行融合得到新视点图像;最后用插值法填充剩余的较小空洞。实验证明,该新方法能获得令人满意的绘制效果。
It is a key technique to generate a virtual view in 3D Video conference. In this paper, a new algorithm based on DIBR and image mergence is proposed. Firstly the depth images are smoothed by an edge dependent filter and the reference images are normalized so as to reduce the big holes and the artificial edges in desired image. Then an arbitrary virtual view is generated by making use of 3D Image Warping. In order to deal with the visibility problem quickly, the pixels in reference image are processed in an occlusion-compatible order. Then two generated images are merged to eliminate the big holes in the desired view. Finally the remaining holes are filled with an interpolation method. Experimental results show that the proposed method can obtain satisfactory rendering results.
出处
《中国图象图形学报》
CSCD
北大核心
2007年第10期1696-1700,共5页
Journal of Image and Graphics
基金
国家自然科学基金项目(60672052和60572127)
上海市重点学科项目(T0102)
上海市曙光计划资助项目(06SG43)