Systems and methods for text-based personalization of images

摘要

As set forth herein, a computer-implemented method is employed to place personalized text into an image. A location and region within the image is determined where the text is to be placed. The 3D geometry of the surface is estimated proximate to the location where the text is to be placed. A personalized text string is received. The personalized text string is incorporated into the image to appear as if it is rendered onto the surface or object according to the estimated 3D geometry.

主权项

1. A computer-implemented method for placing a personalized text into an image to generate a natural text-rendered image comprising:
determining a location and region within the image where the personalized text is to be placed; estimating a 3D geometry of a planar surface proximate to the location and region where the personalized text is to be placed; receiving a personalized text string; incorporating the personalized text string into the image to appear as if it is rendered onto the planar surface according to the estimated 3D geometry; and outputting the natural text-rendered image; wherein estimating the 3D geometry comprises:
determining a first convergent line set and a second convergent line set, wherein the lines within each set are deemed to be parallel in the image;identifying a first vanishing point and a second vanishing point based at least in part upon the first and second convergent line sets;determining a field of view of the image; anddetermining a normal vector of the planar surface; wherein incorporating the personalized text string into the image comprises rendering the personalized text string into the determined location and region within the image using a text mask, wherein the text mask is used to modulate one or more of texture, luminance, chrominance, and transparency of the image; and wherein texture modulation comprises:
utilizing a priori a plurality of images to derive sets of resolution independent basis functions that describe real world textures under a range of illumination conditions and viewpoints;fitting a texture within the determined location and region of a selected image to a linear combination of the basis functions to generate basis weights; andmodulating the basis weights in real time to impose a modulation on the texture in a natural way that creates readable text within the image.