Unlock instant, AI-driven research and patent intelligence for your innovation.

An Adaptive Depth Image Filtering Method

A depth map, adaptive technology, applied in the field of 3D video processing, can solve the problem of inaccurate motion estimation

Active Publication Date: 2017-03-15
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

The existing technology usually only refers to the texture information in the process of pixel motion estimation, but the depth information also reflects the motion state of the object, resulting in inaccurate motion estimation

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • An Adaptive Depth Image Filtering Method
  • An Adaptive Depth Image Filtering Method
  • An Adaptive Depth Image Filtering Method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0055] Such as figure 1 Shown:

[0056] Select the standard test sequence "ballet" for the experiment.

[0057] First, the 3-viewpoint depth map is compressed with HEVC 13.0, the QP is 40, the loop filter is turned off, and the texture map is not processed.

[0058] S1. Define the airspace reference depth map and the time domain reference depth map. The specific steps are as follows:

[0059] Let the current depth image to be filtered be the depth image at the third viewpoint time 3 (the third frame), define the depth image corresponding to the current moment t=3 as the spatial reference depth image, and define the depth image corresponding to the time m as the temporal reference depth image, where m=2,4;

[0060] S2. Perform the spatial reference depth map, the temporal reference depth map, the texture map corresponding to the spatial reference depth map, and the texture map corresponding to the temporal reference depth map described in S1. The mirror image extension of ...

Embodiment 2

[0083] Such as figure 1 Shown:

[0084] Select the standard test sequence "ballet" for the experiment.

[0085] The 5-viewpoint depth map is compressed with HEVC 13.0, the QP is 40, the loop filter is turned off, and the texture map is not processed.

[0086] S1. Define the airspace reference depth map and the time domain reference depth map. The specific steps are as follows:

[0087] Let the current depth image to be filtered be the depth image at the 5th viewpoint time 6 (frame 6), define the depth image corresponding to the current moment t=6 as the spatial reference depth image, and define the depth image corresponding to the time m as the temporal reference depth image, Among them, m=4,5,7,8;

[0088] S2. Perform the spatial reference depth map, the temporal reference depth map, the texture map corresponding to the spatial reference depth map, and the texture map corresponding to the temporal reference depth map described in S1. mirror extension of pixels. The spat...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

PUM

No PUM Login to View More

Abstract

The invention belongs to the technical field of three-dimensional video processing, and particularly relates to a depth-based filtering method of a depth map in a three-dimensional video. The present invention finds the reference pixels in the depth map to be filtered and the neighborhood depth map, uses the pixel vector to predict the motion of the pixels to be filtered, and selects the reference pixels according to the prediction result. Among the reference pixels, pixels that are very related to the pixels to be filtered are selected. Different methods are used in the process of screening pixels that are very related to the pixels to be filtered for smooth regions and border regions, which can more accurately screen out pixels that ultimately participate in filtering. The invention improves the continuity in the time domain direction of the depth map at each moment after filtering. Using different methods for the smooth area and the boundary area can make the smooth area smoother, while maintaining the boundary characteristics of the depth map, and has a certain depth map boundary correction ability, which makes the edge area of ​​the composite image The distortion is significantly reduced, and the quality is improved. Significantly improved.

Description

technical field [0001] The invention belongs to the technical field of three-dimensional video processing, and in particular relates to a filtering method of a depth map in a depth-based three-dimensional video. Background technique [0002] The depth-based 3D video content consists of a texture map and a depth map. The texture map intuitively presents the texture characteristics of the object, and the depth map reflects the distance between the object and the camera. Using the rendering technology based on the depth image, the specified virtual view can be synthesized. texture image. Among them, as the important information of virtual view synthesis, the depth map often introduces a lot of noise in the process of acquisition, compression and transmission, and these noises will cause serious geometric distortion during the synthesis process. The depth image filtering method can effectively remove the noise in the depth image, thereby improving the quality of virtual view sy...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More

Application Information

Patent Timeline
no application Login to View More
Patent Type & Authority Patents(China)
IPC IPC(8): H04N13/00
Inventor 朱策王昕周益民付闻浩程洪
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA