Spatial domain complexity reduction method for depth image based rendering using wavelet transform

Depth Image Based Rendering (DIBR) is an approach to generate a 3-D image by the original 2-D color image with the corresponding 2-D depth map. Although DIBR is a quite convenient technique of converting 2D to 3D images, there is a big problem in DIBR system that it cannot reach real-time processing due to the computing time. Therefore, this paper proposes a method based on discrete wavelet transform and adaptive edge-oriented smoothing process to improve the computing time of the system. The proposed method also preserves the original texture. As a results, this indicate that the proposed method not only preserves the vertical texture but also reduces at least 60% of the computing time of the DIBR system.

I. INTRODUCTION

The three-dimensional (3-D) images have become more and more popular in our life. It provides users with a feeling of presence of 3-D view from simulation of reality. Generally, stereoscopic images must be synthesized by at least two or more 2-D images such that it needs to take more than two times of transmission time and storage space in the coding and transmission stage. European IST started the ATTEST project [1], in which they proposed a Depth Image Based Rendering (DIBR) technology, such that the image for 3-D synthesis can be generated from a single 2-D image and the corresponding depth map to improve the deficiency of the traditional 3-D TV. The DIBR technology not only reduces the transmission time but also reduces storage space of images. There are some factors to affect the quality of DIBR: 1) Occlusion and disocclusion generated by image warping; 2) Imperfect depth map; 3) It cannot preserve the completed horizontal and vertical texture. Occlusion will happen after image warping because the background pixels and foreground pixels may overlap. It can use the foreground pixels to replace the background pixels [2] to resolve the occlusion. Disocclusion usually happens on the edge of the foreground and background of the depth map. It will appear in the newly exposed areas (so called “holes”) in the virtual image after warping. Several approaches were proposed to fill the holes, such as interpolation [3], extrapolation [3], mirroring [4], and image inpainting [4]. Chen et al. [5] proposed an approach by using a smoothing filter on the depth map to reduce the holes. However, this approach is not good to preserve the vertical texture. For solving this problem, Zhang et al. [7] proposed using the asymmetric smoothing filters to preserve the vertical texture. However, the use of asymmetric smoothing filters to process the whole depth map, the horizontal texture may be destroyed. Therefore, Lee et al. [8] proposed a smoothing filter based on the analysis of vertical texture to overcome the difficulties. Tam et al. used the symmetric smoothing filter to process the weaker vertical texture and used the asymmetric smoothing filter [7] to process the stronger vertical texture, and by such arrangement the disocclusion artifacts were incrementally removed as the smoothing of depth maps became stronger. This approach can reduce the geometric distortions as well.