公开(公告)号：US20140105505A1

公开(公告)日：2014-04-17

申请号：US13832122

申请日：2013-03-15

Applicant: GOOGLE INC.

Inventor： Sergey Ioffe ,  Mohamed Aly ,  Charles J. Rosenberg

IPC: G06F17/30

CPC classification number: G06F17/30247 ,  G06F17/3025 ,  G06K9/4676 ,  G06K9/6202

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for determining image search results. One of the methods includes generating a plurality of feature vectors for each image in a collection of images, wherein each feature vector is associated with an image tile of an image, wherein each feature vector corresponds to one of a plurality of predetermined visual words. All images in the collection of images that share at least a threshold number of matching visual words associated with matching image tiles are classified as near-duplicate images.

Abstract translation: 方法，系统和装置，包括在计算机存储介质上编码的计算机程序，用于确定图像搜索结果。 一种方法包括为图像集合中的每个图像生成多个特征向量，其中每个特征向量与图像的图像块相关联，其中每个特征向量对应于多个预定视觉词中的一个。 共享与匹配的图像块相关联的至少阈值数量的匹配视觉词的图像集合中的所有图像被分类为近似重复的图像。

公开(公告)号：US09063954B2

公开(公告)日：2015-06-23

申请号：US13832122

申请日：2013-03-15

Applicant: Google Inc.

Inventor： Sergey Ioffe ,  Mohamed Aly ,  Charles J. Rosenberg

IPC: G06F17/30 ,  G06K9/46 ,  G06K9/62

CPC classification number: G06F17/30247 ,  G06F17/3025 ,  G06K9/4676 ,  G06K9/6202

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for determining image search results. One of the methods includes generating a plurality of feature vectors for each image in a collection of images, wherein each feature vector is associated with an image tile of an image, wherein each feature vector corresponds to one of a plurality of predetermined visual words. All images in the collection of images that share at least a threshold number of matching visual words associated with matching image tiles are classified as near-duplicate images.

Abstract translation: 方法，系统和装置，包括在计算机存储介质上编码的计算机程序，用于确定图像搜索结果。 一种方法包括为图像集合中的每个图像生成多个特征向量，其中每个特征向量与图像的图像块相关联，其中每个特征向量对应于多个预定视觉词中的一个。 共享与匹配的图像块相关联的至少阈值数量的匹配视觉词的图像集合中的所有图像被分类为近似重复的图像。

公开(公告)号：US09189853B1

公开(公告)日：2015-11-17

申请号：US14306757

申请日：2014-06-17

Applicant: Google Inc.

Inventor： Mohamed Aly ,  Jean-Yves Bouguet

IPC: G06T7/00 ,  G06K9/00

CPC classification number: G06T7/0038 ,  G06K9/00664 ,  G06T7/38 ,  G06T7/73 ,  G06T2207/10016

Abstract: Methods and systems for automatically generating pose estimates from uncalibrated unordered panoramas are provided. An exemplary method of automatically generating pose estimates includes receiving a plurality of uncalibrated and unordered panoramic images that include at least one interior building image, and extracting, for each panoramic image, feature points. The method includes generating a match matrix for all the panoramic images based on the one or more feature points, constructing a minimal spanning tree based on the match matrix, identifying a first and second panoramic image, based on the minimal spanning tree, wherein the second panoramic image is associated with the first panoramic image providing a navigation from the first panoramic image to the second panoramic image.

Abstract translation: 提供了用于自动产生未校准的无序全景的姿势估计的方法和系统。 自动产生姿势估计的示例性方法包括：接收多个未校准和无序的全景图像，其包括至少一个室内建筑图像，并且为每个全景图像提取特征点。 该方法包括基于一个或多个特征点为所有全景图像生成匹配矩阵，基于匹配矩阵构建最小生成树，基于最小生成树识别第一和第二全景图像，其中第二个 全景图像与提供从第一全景图像到第二全景图像的导航的第一全景图像相关联。