Image Virtual Viewpoint Generation Method under Hole Pixel Information Update

A virtual viewpoint generation method is proposed to address the problem of low fidelity in viewpoints for images with overlapping pixel points. Virtual factors such as overlaps, holes, cracks, and artifacts are analyzed preprocessed. When background hole a simple texture, pheromone information around used support, at edge detected, predicted same time, so that area filled blocks. has relatively complex depth pixels updated inverse 3D transformation method, projected onto auxiliary plane compared known parameters. The filling performed according symmetry position reference obtain after optimization. was validated using image quality metrics objective evaluation PSNR. experimental results show could generate high fidelity, excellent quality, short image-processing which effectively enhanced performance.