Method for drawing virtual view color image

Abstract

The invention discloses a method for drawing a virtual view color image. In the method, a depth image is divided into a core content area and a non-core content area according to influence of depth coding distortion of different areas on drawing of the virtual view image, and two groups of bidirectional filters with different filtering intensities are designed to perform filter processing on a scene depth value corresponding to each pixel point of the core content area and the non-core content area respectively, so subjective quality of drawing the image is greatly improved; and then reference color information irrelative to void is extracted from the virtual view color image for color distortion problem during image fusion for depth image-based drawing, and different color correction methods are used for correcting colors of two groups of virtual view color images, so the subjective quality of drawing the image is improved.

Description

technical field [0001] The invention relates to a three-dimensional video technology, in particular to a color image rendering method of a virtual viewpoint. Background technique [0002] Three-dimensional video (Three-Dimensional Video, 3DV) is an advanced visual mode, which enables people to have a sense of three-dimensionality and immersion when viewing images on a screen, and can meet people's needs for viewing three-dimensional (3D) scenes from different angles. Typically, 3D video systems such as figure 1 As shown, it mainly includes modules such as video capture, video encoding, transmission decoding, virtual viewpoint rendering and interactive display. [0003] Multi-view video plus depth (MVD) is currently recommended by ISO / MPEG to represent 3D scene information. MVD data adds the depth information of corresponding viewpoints on the basis of multi-viewpoint color images. There are currently two basic ways to obtain depth information: 1) through depth cameras; 2) ...

AI Technical Summary

Problems solved by technology

[0004] The current DIBR method focuses on optimizing the algorithm (such as: how to accurately fill the hole pixels, how to reduce the time of three-dimensional image transformation, etc.) to improve the accuracy and speed of drawing, but for the carrier image (color image and depth image) The effect of rendering quality is lack of relevant research

Through the feature analysis of color images and depth images, on the one hand, since depth is used to represent the load information of scene geometry, the quality of depth information will have an impact on the virtual viewpoint rendering in the later stage. Due to the coding distortion of depth images, the virtual viewpoint drawn There will be geometric distortion (also known as structural position distortion) between the image and the real image, which will generate new holes in the drawn virtual viewpoint image, and the coding distortion and geometric distortion of the depth image are not a simple linear mapping relationship; another On the one hand, because multi-viewpoint imaging will lead to inconsistencies in the color of the collected multi-viewpoint color images, color distortion will occur during the image fusion process of the DIBR method, which will seriously affect the subjective quality of the drawn image

Images