公开(公告)号：US09373190B2

公开(公告)日：2016-06-21

申请号：US14327094

申请日：2014-07-09

Applicant: Google Inc.

Inventor： Xiaolin Wei ,  Aleksey Golovinskiy

IPC: G06T17/05

CPC classification number: G06T17/05 ,  G06T7/33 ,  G06T7/593 ,  G06T2207/10028 ,  G06T2207/30184 ,  G06T2219/2004

Abstract: 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.

Abstract translation: 提供了一种具有深度图对齐和异常值识别的新颖的立体声重建流水线。 一个示例性方法包括获得描绘场景的多个图像。 该方法包括确定多个图像中的每个图像的姿态。 该方法包括确定多个图像中的每一个的深度图，以便确定多个深度图。 多个深度图中的每一个描述与三维空间中对应于场景中的对象的多个点。 该方法包括通过变换多个深度图中的一个或多个来对准多个深度图，以便改善多个深度图之间的对准。 该方法包括识别一个或多个外围点。 该方法包括至少部分地基于多个深度图来生成场景的三维模型。

公开(公告)号：US20150294477A1

公开(公告)日：2015-10-15

申请号：US14248412

申请日：2014-04-09

Applicant: Google Inc.

Inventor： Xiaolin Wei ,  Hui Lou

IPC: G06T7/00 ,  G06K9/48

CPC classification number: G06T7/0083 ,  G06K9/00637 ,  G06K9/4609 ,  G06K9/48 ,  G06T7/12 ,  G06T2207/10032 ,  G06T2207/20036 ,  G06T2207/20061 ,  G06T2207/30184

Abstract: Systems and methods for image-based bridge identification and boundary detection are provided. One example method includes determining, by one or more computing devices, a wide bridge mask and a narrow bridge mask based at least in part on a road mask and a water mask. The method includes selecting, by the one or more computing devices, a portion of an image depicting a bridge based at least in part on the wide bridge mask and the narrow bridge mask. The method includes identifying, by the one or more computing devices, a plurality of line features included in the portion of the image depicting the bridge. The method includes determining, by the one or more computing devices, a bridge boundary in the portion of the image based at least in part on the plurality of line features. One example system includes a bridge image extraction module and a bridge boundary determination module.

Abstract translation: 提供了基于图像的桥梁识别和边界检测的系统和方法。 一个示例性方法包括至少部分地基于道路掩模和水掩模，由一个或多个计算设备确定宽桥接掩模和窄桥接掩模。 所述方法包括至少部分地基于所述宽桥接掩码和所述窄桥接掩码来由所述一个或多个计算设备选择描绘桥接器的图像的一部分。 该方法包括由一个或多个计算设备识别包括在描绘桥的图像的部分中的多个线特征。 所述方法包括至少部分地基于所述多个线特征来由所述一个或多个计算设备确定所述图像的所述部分中的桥接边界。 一个示例系统包括桥接图像提取模块和桥接边界确定模块。

公开(公告)号：US09196050B2

公开(公告)日：2015-11-24

申请号：US14248412

申请日：2014-04-09

Applicant: Google Inc.

Inventor： Xiaolin Wei ,  Hui Lou

IPC: G06K9/48 ,  G06T7/00

CPC classification number: G06T7/0083 ,  G06K9/00637 ,  G06K9/4609 ,  G06K9/48 ,  G06T7/12 ,  G06T2207/10032 ,  G06T2207/20036 ,  G06T2207/20061 ,  G06T2207/30184

Abstract: Systems and methods for image-based bridge identification and boundary detection are provided. One example method includes determining, by one or more computing devices, a wide bridge mask and a narrow bridge mask based at least in part on a road mask and a water mask. The method includes selecting, by the one or more computing devices, a portion of an image depicting a bridge based at least in part on the wide bridge mask and the narrow bridge mask. The method includes identifying, by the one or more computing devices, a plurality of line features included in the portion of the image depicting the bridge. The method includes determining, by the one or more computing devices, a bridge boundary in the portion of the image based at least in part on the plurality of line features. One example system includes a bridge image extraction module and a bridge boundary determination module.

Abstract translation: 提供了基于图像的桥梁识别和边界检测的系统和方法。 一个示例性方法包括至少部分地基于道路掩模和水掩模，由一个或多个计算设备确定宽桥接掩模和窄桥接掩模。 所述方法包括至少部分地基于所述宽桥接掩码和所述窄桥接掩码来由所述一个或多个计算设备选择描绘桥接器的图像的一部分。 该方法包括由一个或多个计算设备识别包括在描绘桥的图像的部分中的多个线特征。 所述方法包括至少部分地基于所述多个线特征来由所述一个或多个计算设备确定所述图像的所述部分中的桥接边界。 一个示例系统包括桥接图像提取模块和桥接边界确定模块。

公开(公告)号：US09373192B2

公开(公告)日：2016-06-21

申请号：US13915653

申请日：2013-06-12

Applicant: Google Inc.

Inventor： Xiaolin Wei

IPC: G06T17/20

CPC classification number: G06T17/205

Abstract: Systems and methods for simplifying a three-dimensional mesh providing a three-dimensional model, such as a three-dimensional model of a cityscape, are provided. In particular, mesh simplification techniques can preserve important structural shapes in the original three-dimensional mesh. Important features in a representation of a city scene or other scene can be represented by structural edges (e.g. building outlines, bridges, curved structures such as curved highway overpasses, etc.). These structural edges can be detected and a representation of the structural edges can be constructed in the mesh. The mesh can then be simplified using the structural edges as high-level simplification criterion such that the structural edges are preserved during simplification.

Abstract translation: 提供了用于简化提供三维模型（例如城市景观的三维模型）的三维网格的系统和方法。 特别地，网格简化技术可以保留原始三维网格中重要的结构形状。 城市场景或其他场景的表示中的重要特征可以由结构边缘（例如建筑物轮廓，桥梁，诸如弯曲的公路立交桥等的弯曲结构等）来表示。 可以检测这些结构边缘，并且可以在网格中构造结构边缘的表示。 然后可以使用结构边缘作为高级简化标准来简化网格，使得在简化期间保留结构边缘。

公开(公告)号：US09437034B1

公开(公告)日：2016-09-06

申请号：US14569833

申请日：2014-12-15

Applicant: Google Inc.

Inventor： Xiaolin Wei ,  Tilman Reinhardt

IPC: G06T15/00 ,  G06T15/04 ,  G06T17/05 ,  G06T17/00 ,  G06T3/40

CPC classification number: G06T15/04 ,  G06T3/40 ,  G06T17/00

Abstract: Systems and methods for generating textures to be rendered in conjunction with a polygon mesh are provided. More particularly, a polygon mesh modeling a geographic area can be accessed. A plurality of source images depicting the geographic area can then be identified. The plurality of source images can be aligned to reduce projection misalignments between source images when the source images are projected to the polygon mesh. A texture can then be determined based at least in part on a weighted average of pixels in the plurality of source images corresponding to a point on the surface of the polygon mesh. Determining a texture can include removing at least one outlier pixel associated with a moving object from the weighted average.

Abstract translation: 提供了用于生成要与多边形网格一起渲染的纹理的系统和方法。 更具体地，可以访问对地理区域建模的多边形网格。 然后可以识别描绘地理区域的多个源图像。 当源图像投影到多边形网格时，可以对准多个源图像以减少源图像之间的投影不对准。 然后可以至少部分地基于对应于多边形网格的表面上的点的多个源图像中的像素的加权平均来确定纹理。 确定纹理可以包括从加权平均值移除与移动对象相关联的至少一个异常值像素。

公开(公告)号：US20140368498A1

公开(公告)日：2014-12-18

申请号：US13915653

申请日：2013-06-12

Applicant: Google Inc.

Inventor： Xiaolin Wei

IPC: G06T19/00

CPC classification number: G06T17/205

Abstract: Systems and methods for simplifying a three-dimensional mesh providing a three-dimensional model, such as a three-dimensional model of a cityscape, are provided. In particular, mesh simplification techniques can preserve important structural shapes in the original three-dimensional mesh. Important features in a representation of a city scene or other scene can be represented by structural edges (e.g. building outlines, bridges, curved structures such as curved highway overpasses, etc.). These structural edges can be detected and a representation of the structural edges can be constructed in the mesh. The mesh can then be simplified using the structural edges as high-level simplification criterion such that the structural edges are preserved during simplification.

Abstract translation: 提供了用于简化提供三维模型（例如城市景观的三维模型）的三维网格的系统和方法。 特别地，网格简化技术可以保留原始三维网格中重要的结构形状。 城市场景或其他场景的表示中的重要特征可以由结构边缘（例如建筑物轮廓，桥梁，诸如弯曲的公路立交桥等的弯曲结构等）来表示。 可以检测这些结构边缘，并且可以在网格中构造结构边缘的表示。 然后可以使用结构边缘作为高级简化标准来简化网格，使得在简化期间保留结构边缘。

公开(公告)号：US20160012633A1

公开(公告)日：2016-01-14

申请号：US14327094

申请日：2014-07-09

Applicant: Google Inc.

Inventor： Xiaolin Wei ,  Aleksey Golovinskiy

IPC: G06T17/05

CPC classification number: G06T17/05 ,  G06T7/33 ,  G06T7/593 ,  G06T2207/10028 ,  G06T2207/30184 ,  G06T2219/2004

Abstract: 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.

Abstract translation: 提供了一种具有深度图对齐和异常值识别的新颖的立体声重建流水线。 一个示例性方法包括获得描绘场景的多个图像。 该方法包括确定多个图像中的每个图像的姿态。 所述方法包括确定所述多个图像中的每一个的深度图，使得确定多个深度图。所述多个深度图中的每个深度图描述与所述场景中的对象相对应的三维空间中的多个点。 该方法包括通过变换多个深度图中的一个或多个来对准多个深度图，以便改善多个深度图之间的对准。 该方法包括识别一个或多个外围点。 该方法包括至少部分地基于多个深度图来生成场景的三维模型。