Stereoscopic image redirecting method

Abstract

The invention relates to a stereoscopic image redirecting method which comprises: firstly establishing an energy equation according to the texture similarity and parallax similarity of pixels; establishing a cost matrix M according to the established energy equation and selecting a line with the lowest energy cost by dynamic programming; combining depth information with a plane saliency map to obtain a stereoscopic saliency map; performing spatial coherence filtering on the saliency map to obtain a final importance image IM; allocating a scaling factor to a left view; finding out a corresponding relation between the pixels in the left view and a right view according to theparallax of left and right views, and obtaining the scaling factor of the pixels of the right view; according to the scaling factor of each, redirecting the image by using a pixel fusion algorithm to obtain a final redirected image. The stereoscopic image redirecting method can reduce visual distortion.

Description

Technical field [0001] The invention belongs to the technical fields of image processing and stereo vision, and relates to a method for reorienting stereo images capable of saving depth. Background technique [0002] In order to apply images to different systems and display them optimally, it is often necessary to modify the resolution of the image. This process is called image redirection. The goal is to modify the resolution of a given image, enlarge or reduce the image while maintaining important features in the image and minimizing its distortion. Compared with traditional planar images, the depth information of stereo images can bring valuable clues for redirection. However, the preservation of this information also brings new challenges for stereo image redirection. Stereoscopic image redirection must not only save important areas in the image, reduce visual distortion, but also reduce parallax distortion. The change of the disparity value will affect the stereo percepti...

AI Technical Summary

Problems solved by technology

Since the idea of ​​the algorithm is to embed the image into the grid for deformation so as to control the image content, when the unimportant content is stretched to different degrees, the global structure of the image may be destroyed

At the same time, discontinuous deformation and parallax distortion may occur due to the disparity preservation of only some discrete points rather than all points

Considering that although the pixel removal method is more flexible, but the ability to preserve the object structure is weak, Yan et al. used pixel fusion instead of pixel removal to better solve the problem of zigzag borders caused by line clipping of planar images.

Images