Wide-angle portrait picture distortion correction method based on adaptive grid

Abstract

The invention discloses a wide-angle portrait picture distortion correction method based on an adaptive grid, and relates to the field of image deformation. The method includes the following steps: 1) inputting a wide-angle perspective image, and setting related parameters; 2) calculating a corresponding portrait significance density map; 3) generating self-adaptive triangular mesh generation according to the density map; 4) calculating new positions of vertexes of the triangular mesh according to an energy function for measuring distortion provided by the invention, and moving the vertexes to the corresponding new positions; and 5) driving the image to deform according to the movement of the triangular mesh, and re-rendering the image to obtain a wide-angle image of which the edge figure has no obvious deformation. By using the strong adaptability of the triangular mesh, the number of vertexes and patches of the triangular mesh can be reduced to the greatest extent, the distortion minimization efficiency is improved, and a good visual effect is achieved. The energy function design has high expansibility, specific geometric constraints can be added, and personal requirements of users are met through proper interaction. The problem of portrait distortion at the edge of a wide-angle photo is solved.

Description

technical field [0001] The invention relates to the field of image deformation, in particular to an adaptive grid-based wide-angle portrait photo distortion correction method. Background technique [0002] Wide-angle portrait photo distortion is an important direction in the field of image deformation. The general solution is to combine the imaging effects produced by different projection imaging models to correct the image in different regions to maintain the natural portrait and background effects. [0003] The combination of perspective projection (Perspective Projection) and stereographic projection (Stereographic Projection) or Mercator projection (Mercator Projection) is a mainstream combination method. Perspective projection has the characteristic of maintaining the straight line structure in the camera lens, but the disadvantage is that when the camera When the viewing angle is large, there will be obvious stretching at the edge of the picture, resulting in unnatural...

AI Technical Summary

Problems solved by technology

But in fact, for non-salient areas, there is no need for too dense grids to drive deformation, and uniform grid distribution will make non-salient areas also have dense vertex distribution. Using too many vertices will lead to the optimization problem of minimizing distortion. consume more time in

Images