No abstract

A high quality DIBR based free viewpoint generation model is proposed in this paper. Firstly, a boundary-aware 3D warping is introduced to project the side views from the side view plane to the virtual view plane. During the warping, a high-pass (HP) filter is used to remove the ghost contours around the object boundaries. Secondly, a color correction based image blending (CC-IB) procedure is put forward for synthesizing the two warped side views. In this CC-IB process, the disoccluded regions are color-corrected in the first place, based on an equivalent substitution and a polynomial fitting method. By blending the color-corrected disocclusions and the un-occluded regions, a color-continuous blended virtual view is produced. Lastly, a depth guided hierarchical hole-filling (D-HHF) is presented to fill the holes in the blended virtual view. With the assistance of depth, holes are divided into two kinds and only background pixels will be selected for filling the holes. The proposed D-HHF not only results in clear textures but also executes in a fast speed. Experimental results show that the proposed method can produce high quality virtual views with comfortable visual perceptions. Comparisons with the other methods demonstrate that the proposed method can achieve good performance in both visual quality and quantitative evaluation.

Depth-image-based-rendering (DIBR) has received much attention in recent years as a promising technology for 3DTV systems. However, holes are inevitable in DIBR during the view synthesis procedure because the scene area, which has been occluded in the reference image, become visible in the synthesized virtual view. In this paper, we propose dictionary based hole filling with the assistance of depth. Because holes generally come from background area, we first segment background from foreground with the assistance of depth. Then, we construct a dictionary for hole filling from background. Finally, we employ the dictionary to fill holes in the synthesized virtual view. Experimental results demonstrate that the proposed method achieves good performance in hole filling in terms of visual quality and quantitative measures.

No abstract

A multi-scale simple linear iterative clustering (SLIC)-based depth up-sampling method is proposed in order to obtain high-quality depth maps, especially in the case of high up-sampling rate. The proposed method is implemented hierarchically, where the high-resolution image is segmented from coarse to fine by using multi-scale SLIC superpixels. A depth guided discriminant function is defined to distinguish the validity of the segmented superpixels, and only the valid ones will be interpolated each layer. The experimental results show that the proposed method solves the depth missing and the depth confusion problems largely.