利索-全球专利检索

  1. 登录
  2. 注册

搜索结果

26 条记录 (耗时 0.052 秒)

    PLANAR SURFACE DETECTION
    5.
    发明申请
    PLANAR SURFACE DETECTION 有权
    平面表面检测

    公开(公告)号：US20140118397A1

    公开(公告)日：2014-05-01

    申请号：US13660948

    申请日：2012-10-25

    申请人： Kyungsuk David Lee ,  Alexandru Balan ,  Jeffrey N. Margolis

    发明人： Kyungsuk David Lee ,  Alexandru Balan ,  Jeffrey N. Margolis

    IPC分类号： G06K9/62 ,  G09G5/00

    CPC分类号： G06T7/0057 ,  G06T7/521 ,  G06T2207/10028

    摘要： A planar surface within a physical environment is detected enabling presentation of a graphical user interface overlaying the planar surface. Detection of planar surfaces may be performed, in one example, by obtaining a collection of three-dimensional surface points of a physical environment imaged via an optical sensor subsystem. A plurality of polygon sets of points are sampled within the collection. Each polygon set of points includes three or more localized points of the collection that defines a polygon. Each polygon is classified into one or more groups of polygons having a shared planar characteristic with each other polygon of that group. One or more planar surfaces within the collection are identified such that each planar surface is at least partially defined by a group of polygons containing at least a threshold number of polygons.

    摘要翻译： 检测物理环境内的平坦表面，使得能够呈现覆盖平面表面的图形用户界面。 在一个示例中，可以通过获得经由光学传感器子系统成像的物理环境的三维表面点的集合来执行平面表面的检测。 在集合内采样多个多边形集合点。 每个多边形集合点包括定义多边形的集合的三个或更多个局部点。 每个多边形被分成具有与该组的每个其他多边形共享的平面特征的一组或多组多边形。 识别集合内的一个或多个平面表面，使得每个平面表面至少部分地由包含至少一个阈值数量的多边形的一组多边形限定。

    Fully automatic dynamic articulated model calibration
    6.
    发明授权
    Fully automatic dynamic articulated model calibration 有权
    全自动动态关节模型校准

    公开(公告)号：US08610723B2

    公开(公告)日：2013-12-17

    申请号：US13172255

    申请日：2011-06-29

    申请人： Kyungsuk David Lee ,  Alexandru Balan

    发明人： Kyungsuk David Lee ,  Alexandru Balan

    IPC分类号： G06T13/00

    CPC分类号： G06T17/00 ,  G06T13/40

    摘要： A depth sensor obtains images of articulated portions of a user's body such as the hand. A predefined model of the articulated body portions is provided. The model is matched to corresponding depth pixels which are obtained from the depth sensor, to provide an initial match. The initial match is then refined using distance constraints, collision constraints, angle constraints and a pixel comparison using a rasterized model. Distance constraints include constraints on distances between the articulated portions of the hand. Collision constraints can be enforced when the model meets specified conditions, such as when at least two adjacent finger segments of the model are determined to be in a specified relative position, e.g., parallel. The rasterized model includes depth pixels of the model which are compared to identify overlapping pixels. Dimension of the articulated portions of the model are individually adjusted.

    摘要翻译： 深度传感器获得诸如手的用户身体的关节部分的图像。 提供了铰接体部分的预定模型。 该模型与从深度传感器获得的相应深度像素匹配，以提供初始匹配。 然后使用距离约束，碰撞约束，角度约束和使用光栅化模型的像素比较来精简初始匹配。 距离约束包括对手的铰接部分之间距离的约束。 当模型满足指定的条件时，例如当模型的至少两个相邻手指段被确定为处于指定的相对位置（例如并行）时，可以强制执行冲突约束。 光栅化模型包括与识别重叠像素进行比较的模型的深度像素。 模型的铰接部分的尺寸被单独调整。

    Probabilistic And Constraint Based Articulated Model Fitting
    7.
    发明申请
    Probabilistic And Constraint Based Articulated Model Fitting 有权
    基于概率和约束的铰接模型拟合

    公开(公告)号：US20130329011A1

    公开(公告)日：2013-12-12

    申请号：US13688064

    申请日：2012-11-28

    申请人： Kyungsuk David Lee ,  Alexandru Balan

    发明人： Kyungsuk David Lee ,  Alexandru Balan

    IPC分类号： H04N13/02 ,  G06T17/00

    CPC分类号： H04N13/275 ,  G06F3/011 ,  G06F3/017 ,  G06T7/75 ,  G06T17/00 ,  G06T19/20 ,  G06T2200/04 ,  G06T2207/10028 ,  G06T2207/30196 ,  G06T2219/2016 ,  H04N13/271

    摘要： A depth sensor obtains images of articulated portions of a user's body such as the hand. A predefined model of the articulated body portions is provided. Representative attract points of the model are matched to centroids of the depth sensor data, and a rigid transform of the model is performed, in an initial, relatively coarse matching process. This matching process is then refined in a non-rigid transform of the model, using attract point-to-centroid matching. In a further refinement, an iterative process rasterizes the model to provide depth pixels of the model, and compares the depth pixels of the model to the depth pixels of the depth sensor. The refinement is guided by whether the depth pixels of the model are overlapping or non-overlapping with the depth pixels of the depth sensor. Collision, distance and angle constraints are also imposed on the model.

    摘要翻译： 深度传感器获得诸如手的用户身体的关节部分的图像。 提供了铰接体部分的预定模型。 模型的代表性吸引点与深度传感器数据的质心相匹配，并且在初始的较粗略的匹配过程中执行模型的刚性变换。 然后，使用吸引点到质心匹配，在模型的非刚性变换中对该匹配过程进行细化。 在进一步的细化中，迭代过程光栅化模型以提供模型的深度像素，并将模型的深度像素与深度传感器的深度像素进行比较。 细化是由模型的深度像素是否与深度传感器的深度像素重叠或不重叠来指导。 对模型也施加了碰撞，距离和角度限制。

    PARTITIONING DATA FOR TRAINING MACHINE-LEARNING CLASSIFIERS
    9.
    发明申请
    PARTITIONING DATA FOR TRAINING MACHINE-LEARNING CLASSIFIERS 审中-公开
    训练机器学习分类器的分类数据

    公开(公告)号：US20160132786A1

    公开(公告)日：2016-05-12

    申请号：US14539778

    申请日：2014-11-12

    申请人： Alexandru Balan ,  Bradford Jason Snow ,  Christopher Douglas Edmonds ,  Henry Nelson Jerez ,  Kyungsuk David Lee ,  Mark J. Finocchio ,  Miguel Susffalich ,  Cem Keskin

    发明人： Alexandru Balan ,  Bradford Jason Snow ,  Christopher Douglas Edmonds ,  Henry Nelson Jerez ,  Kyungsuk David Lee ,  Mark J. Finocchio ,  Miguel Susffalich ,  Cem Keskin

    IPC分类号： G06N99/00

    CPC分类号： G06N20/00

    摘要： Various embodiments relating to partitioning a data set for training machine-learning classifiers based on an output of a globally trained machine-learning classifier are disclosed. In one embodiment, a first machine-learning classifier may be trained on a set of training data to produce a corresponding set of output data. The set of training data may be partitioned into a plurality of subsets based on the set of output data. Each subset may correspond to a different class. A second machine-learning classifier may be trained on the set of training data using a plurality of classes corresponding to the plurality of subsets to produce, for each data object of the set of training data, a probability distribution having for each class a probability that the data object is a member of the class.

    摘要翻译： 公开了关于根据全球训练的机器学习分类器的输出划分用于训练机器学习分类器的数据集的各种实施例。 在一个实施例中，可以对一组训练数据对第一机器学习分类器进行训练，以产生相应的一组输出数据。 该组训练数据可以基于该组输出数据被划分成多个子集。 每个子集可以对应于不同的类。 可以使用对应于多个子集的多个类来对该组训练数据对第二机器学习分类器进行训练，以针对该组训练数据的每个数据对象产生对于每个类具有以下概率的概率： 数据对象是该类的成员。