Adaptive depth truncation filter for MVC based compressed depth image

作者：

Highlights：

• Boundary artifacts of the synthesized views are minimized by sharpening the depth map edges that are blurred by subsampling process or transform coding of 3DV-ATM.

• Sharp depth edge pixels are adaptively detected with horizontal and vertical gradients.

• Depth image is divided into non-overlapped blocks for edge block detection. To increase the accuracy of depth map refining, each edge block is adaptively repositioned and then expanded in order to locate the edge regions at the center of the block.

• A modified block truncation technique is used to recover the sharp edges within the edge blocks.

摘要

•Boundary artifacts of the synthesized views are minimized by sharpening the depth map edges that are blurred by subsampling process or transform coding of 3DV-ATM.•Sharp depth edge pixels are adaptively detected with horizontal and vertical gradients.•Depth image is divided into non-overlapped blocks for edge block detection. To increase the accuracy of depth map refining, each edge block is adaptively repositioned and then expanded in order to locate the edge regions at the center of the block.•A modified block truncation technique is used to recover the sharp edges within the edge blocks.

论文关键词：Multiview video plus depth (MVD),Depth image based rendering (DIBR),3DV-ATM,Multiview video coding (MVC),Depth image compression,Depth image filter

论文评审过程：Received 2 April 2013, Revised 19 December 2013, Accepted 20 December 2013, Available online 8 January 2014.

论文官网地址：https://doi.org/10.1016/j.image.2013.12.005