Google Inc.

Aleksey Golovinskiy

Brett Allen

Michael Hongmai Lin

Tilman Reinhardt

Scott Edward Dillard

Xiaolin Wei

David Robert Gallup

Alvin Jon-Hang Law

Michael Hongnai Lin

Systems and methods for determining two-dimensional (2D) images are presented. For instance, data indicative of a three-dimensional (3D) model of a geographic area can be obtained. A 2D output image can be generated depicting at least a portion of the geographic area based at least in part on the 3D model. Each pixel in the output image can then be reprojected to the 3D model. A plurality of aerial images depicting the geographic area can be obtained. A source image can then be determined for each pixel in the output image from the plurality of aerial images. The source image can be determined based at least in part on the reprojection of the pixel in the output image to the three-dimensional model.

Determining two-dimensional images using three-dimensional models

Systems and methods for rendering a view-dependent texture in conjunction with a three-dimensional model of a geographic area are provided. A view-dependent texture can be rendered in conjunction with at least portions of the three-dimensional model. A base texture can be rendered for portions of the three-dimensional model in the same field of view that are viewed from a slightly different perspective than a reference direction associated with the view-dependent texture. For instance, a stretching factor can be determined for each portion of the three-dimensional model based on the reference direction and a viewpoint direction associated with the portion of the three-dimensional model. A base texture, a view-dependent texture, or a blended texture can be selected for rendering at the portion of the three-dimensional model based on the stretching factor.

Texture blending between view-dependent texture and base texture in a geographic information system

A novel stereo reconstruction pipeline that features depth map alignment and outlier identification is provided. One example method includes obtaining a plurality of images depicting a scene. The method includes determining a pose for each of the plurality of images. The method includes determining a depth map for each of the plurality of images such that a plurality of depth maps are determined. Each of the plurality of depth maps describes a plurality of points in three-dimensional space that correspond to objects in the scene. The method includes aligning the plurality of depth maps by transforming one or more of the plurality of depth maps so as to improve an alignment between the plurality of depth maps. The method includes identifying one or more outlying points. The method includes generating a three-dimensional model of the scene based at least in part on the plurality of depth maps.

High-quality stereo reconstruction featuring depth map alignment and outlier identification

Systems, methods and computer apparatuses for computing three dimensional (3D) surfaces are described herein. An embodiment includes constructing a slanted voxel grid oriented in a direction between at least two camera locations, projecting images from respective camera locations onto the constructed voxel grid, calculating, for one or more voxels in the voxel grid, a magnitude of difference between pixels corresponding to respective projected images to provide a difference value for each of the one or more voxels, and computing a three dimensional surface passing through voxel locations corresponding to one or more calculated difference values of the one or more voxels. In an embodiment, the three dimensional surface is computed as a surface passing through voxel locations corresponding to calculated difference values having lower magnitudes with respect to other calculated difference values.

Methods and systems to compute 3D surfaces

In an embodiment, a method determines a three-dimensional model from a plurality of images taken of a geographic region by one or more cameras from different perspectives. The method includes determining, using a first stereo reconstruction technique: (i) a plurality of three-dimensional candidate surface points from the plurality of images, and (ii) which of the images in the plurality of images view each of the plurality of candidate surface points. The method also includes identifying an empty space between each of the plurality of candidate surface points and each camera model for the respective images determined to view the candidate surface point. The method further includes for each of a plurality of pairs of images from the plurality of images, determining, using a second stereo reconstruction technique, a surface estimate for the pair of images. The method also includes merging the surface estimates to identify a final surface.

Combining narrow-baseline and wide-baseline stereo for three-dimensional modeling

In an embodiment, a method merges a plurality of three-dimensional models, each having been generated from images of a camera having a different viewpoint. For respective voxels in a plurality of voxels, a plurality of distance values are determined. Each determined distance value is a distance from the voxel to a three dimensional model from the plurality of three-dimensional models along a perspective of a camera model used to generate the three-dimensional model. When at least one of the determined distance values indicates that the distance between the voxel and the three dimensional model exceeds a threshold and the voxel is located above the corresponding three-dimensional model, a confidence score is determined indicating a degree to which the perspective of the camera model of the corresponding three-dimensional model is oriented to face the voxel. Based on the determined confidence score, the voxel into a merged three-dimensional model is determined.

Growing community of inventors

San Francisco, CA, United States of America

Aleksey Golovinskiy