25 September 2013 Deinterlacing of depth-image-based three-dimensional video for a depth-image-based rendering system
Author Affiliations +
Abstract
Wide bandwidth is usually required to transmit depth-image-based three-dimensional (3-D) video. An effective way to reduce the bandwidth requirement is converting the progressive 3-D video to an interlaced one. Since the interlaced video cannot be displayed on an end directly, the video must be deinterlaced. A deinterlacing algorithm for depth-image-based 3-D video is proposed. The algorithm first performs deinterlacing on an interlaced reference image and its interlaced depth map by an intrafield interpolation algorithm based on direction. The interpolation algorithm works well in eliminating jagged effect by adaptively selecting the direction of interpolation; it also provides higher peak SNR in comparison with other commonly used algorithms on interpolating interlaced depth map. Then, 3-D image warping is applied to the progressive reference image and its depth map to generate a progressive destination image. Finally, hole-filling is performed on the destination image to get the virtual view. A method to modify the algorithm to facilitate hardware implementation is also discussed. Experimental results show that the proposed deinterlacing algorithm can provide progressive virtual views with quite good image quality and the performance is robust.
© 2013 SPIE and IS&T
Ran Liu, Weimin Tan, Yingjian Wu, Yingchun Tan, Bole Li, Hui Xie, Guoqin Tai, Xiaoyan Xu, "Deinterlacing of depth-image-based three-dimensional video for a depth-image-based rendering system," Journal of Electronic Imaging 22(3), 033031 (25 September 2013). https://doi.org/10.1117/1.JEI.22.3.033031 . Submission:
JOURNAL ARTICLE
18 PAGES


SHARE
Back to Top