公开(公告)号：EP4476687A1

公开(公告)日：2024-12-18

申请号：EP22823296.3

申请日：2022-11-28

Applicant: QUALCOMM INCORPORATED

Inventor： ZHANG, Shuai ,  YING, Xiaowen ,  LYU, Jiancheng ,  QI, Yingyong

IPC: G06T7/174

公开(公告)号：EP4088260A1

公开(公告)日：2022-11-16

申请号：EP20830473.3

申请日：2020-12-14

Applicant: QUALCOMM INCORPORATED

Inventor： DENG, Yan ,  SARKIS, Michel Adib ,  QI, Yingyong

IPC: G06T7/55

公开(公告)号：EP1751971A2

公开(公告)日：2007-02-14

申请号：EP05713427.2

申请日：2005-02-09

Applicant: Qualcomm, Incorporated

Inventor： QI, Yingyong

IPC: H04N5/228

CPC classification number: H04N5/145 ,  H04N5/23248 ,  H04N5/23254 ,  H04N5/23264

Abstract: A method and apparatus for image stabilization takes an input image sequence including a plurality of frames (23), estimates frame-level motion vectors for each frame, and adaptively integrates the motion vectors to produce, for each frame, a motion vector to be used for image stabilization (7). A copy of the reference image of a frame is displaced by the corresponding adaptively integrated motion vector (21). In one embodiment, the perimeter of the image sensor is padded with margin to be used for image compensation. In another embodiment, vertical and horizontal components are treated independently (7). In still another embodiment, the motion estimation circuitry associated with an MPEG-4 encoder is used to calculate macroblock level vectors (11), and a histogram is used co compute a corresponding frame-level vector for that frame (7).

公开(公告)号：WO2019005291A1

公开(公告)日：2019-01-03

申请号：PCT/US2018/030875

申请日：2018-05-03

Applicant: QUALCOMM INCORPORATED

Inventor： ZHOU, Yang ,  CHEN, Ying ,  QI, Yingyong ,  BI, Ning

IPC: G06T7/246

Abstract: In various implementations, object tracking in a video content analysis system can be augmented with an image-based object re-identification system (e.g., for person re- identification or re-identification of other objects) to improve object tracking results for objects moving in a scene. The object re-identification system can use image recognition principles, which can be enhanced by considering data provided by object trackers that can be output by an object traffic system. In a testing stage, the object re-identification system can selectively test object trackers against object models. For most input video frames, not all object trackers need be tested against all object models. Additionally, different types of object trackers can be tested differently, so that a context provided by each object tracker can be considered. In a training stage, object models can also be selectively updated.

公开(公告)号：WO2018089081A1

公开(公告)日：2018-05-17

申请号：PCT/US2017/048266

申请日：2017-08-23

Applicant: QUALCOMM INCORPORATED

Inventor： SARKIS, Michel Adib ,  QI, Yingyong

IPC: G06K9/00

CPC classification number: G06K9/00288 ,  G06K9/00268 ,  G06K9/6215 ,  G06K9/6288

Abstract: An apparatus includes a processor and a memory. The processor is configured to execute instructions stored at the memory to receive first characterization data and second characterization data. The first characterization data includes first values in a first order and corresponding to a first object. The second characterization data includes second values in a second order and corresponding to a second object. The processor is further configured to generate third characterization data and to generate fourth characterization. The third characterization data includes the first values in a third order. The fourth characterization data includes the second values in a fourth order. The processor is also configured to perform a first similarity operation using the first, second, third, and fourth characterization data to generate first result data and to determine whether the first object and the second object match based on the first result data.

Abstract translation: 一种装置包括处理器和存储器。 处理器被配置为执行存储在存储器处的指令以接收第一表征数据和第二表征数据。 第一特征数据包括第一次序的第一值并且对应于第一对象。 第二表征数据包括以第二顺序并且对应于第二对象的第二值。 该处理器还被配置成生成第三特征数据并生成第四特征。 第三个表征数据包括第三个数值的第一个值。 第四表征数据以第四顺序包括第二值。 处理器还被配置为使用第一，第二，第三和第四表征数据来执行第一相似性操作以生成第一结果数据并且基于第一结果数据来确定第一对象和第二对象是否匹配。

公开(公告)号：WO2017058362A1

公开(公告)日：2017-04-06

申请号：PCT/US2016/045838

申请日：2016-08-05

Applicant: QUALCOMM INCORPORATED

Inventor： GAO, Jinglun ,  GAO, Dashan ,  ZHONG, Xin ,  QI, Yingyong

IPC: H04N5/232

CPC classification number: H04N5/23296 ,  G01S3/00 ,  G01S3/7864 ,  G01S5/16 ,  G06T7/20 ,  G06T7/337 ,  G06T7/74 ,  H04N5/23222 ,  H04N5/23258 ,  H04N5/2328

Abstract: An electronic device is described. The electronic device includes a processor. The processor is configured to obtain a plurality of images. The processor is also configured to obtain global motion information indicating global motion between at least two of the plurality of images. The processor is further configured to obtain object tracking information indicating motion of a tracked object between the at least two of the plurality of images. The processor is additionally configured to perform automatic zoom based on the global motion information and the object tracking information. Performing automatic zoom produces a zoom region including the tracked object. The processor is configured to determine a motion response speed for the zoom region based on a location of the tracked object within the zoom region.

Abstract translation: 描述电子设备。 电子设备包括处理器。 处理器被配置为获得多个图像。 处理器还被配置为获得指示多个图像中的至少两个之间的全局运动的全局运动信息。 处理器还被配置为获得指示所述多个图像中的至少两个图像之间的跟踪对象的运动的对象跟踪信息。 处理器另外被配置为基于全局运动信息和对象跟踪信息执行自动缩放。 执行自动缩放会产生包括跟踪对象的缩放区域。 处理器被配置为基于在缩放区域内被跟踪对象的位置来确定变焦区域的运动响应速度。

公开(公告)号：WO2016032692A1

公开(公告)日：2016-03-03

申请号：PCT/US2015/043313

申请日：2015-07-31

Applicant: QUALCOMM INCORPORATED

Inventor： GAO, Dashan ,  YANG, Yang ,  ZHONG, Xin ,  QI, Yingyong

IPC: G06K9/62

CPC classification number: G06K9/6269 ,  G06K9/4604 ,  G06N7/00 ,  G06N99/005

Abstract: A method for object classification by an electronic device is described. The method includes obtaining an image frame that includes an object. The method also includes determining samples from the image frame. Each of the samples represents a multidimensional feature vector. The method further includes adding the samples to a training set for the image frame. The method additionally includes pruning one or more samples from the training set to produce a pruned training set. One or more non-support vector negative samples are pruned first. One or more non-support vector positive samples are pruned second if necessary to avoid exceeding a sample number threshold. One or more support vector samples are pruned third if necessary to avoid exceeding the sample number threshold. The method also includes updating classifier model weights based on the pruned training set.

Abstract translation: 描述了一种通过电子设备进行物体分类的方法。 该方法包括获得包括对象的图像帧。 该方法还包括从图像帧确定样本。 每个样本表示多维特征向量。 该方法还包括将样本添加到图像帧的训练集合中。 该方法还包括从训练集修剪一个或多个样本以产生修剪的训练集。 首先修剪一个或多个非支持向量负样本。 如果必要，一个或多个非支持向量阳性样本被修剪，以避免超过样本数阈值。 如果需要，一个或多个支持向量样本被修剪为第三，以避免超过样本数阈值。 该方法还包括基于修剪的训练集更新分类器模型权重。

公开(公告)号：WO2012074937A1

公开(公告)日：2012-06-07

申请号：PCT/US2011/062261

申请日：2011-11-28

Applicant: QUALCOMM INCORPORATED ,  ZHANG, Xuerui ,  BI, Ning ,  QI, Yingyong

Inventor： ZHANG, Xuerui ,  BI, Ning ,  QI, Yingyong

IPC: H04N13/00 ,  H04N13/02 ,  G06T19/00

CPC classification number: G06T19/006 ,  H04N13/156

Abstract: A three dimensional (3D) mixed reality system combines a real 3D image or video, captured by a 3D camera for example, with a virtual 3D image rendered by a computer or other machine to render a 3D mixed-reality image or video. A 3D camera can acquire two separate images (a left and a right) of a common scene, and superimpose the two separate images to create a real image with a 3D depth effect. The 3D mixed-reality system can determine a distance to a zero disparity plane for the real 3D image, determine one or more parameters for a projection matrix based on the distance to the zero disparity plane, render a virtual 3D object based on the projection matrix, combine the real image and the virtual 3D object to generate a mixed-reality 3D image.

Abstract translation: 三维（3D）混合现实系统将由3D摄像机捕获的真实3D图像或视频与由计算机或其他机器呈现的虚拟3D图像组合以渲染3D混合现实图像或视频。 3D摄像机可以获取公共场景的两个单独的图像（左和右），并且叠加两个分离的图像以创建具有3D深度效果的实际图像。 3D混合现实系统可以确定实际3D图像到零视差平面的距离，基于到零视差平面的距离确定投影矩阵的一个或多个参数，基于投影矩阵渲染虚拟3D对象 ，组合真实图像和虚拟3D对象以产生混合现实的3D图像。

公开(公告)号：WO2008019262A2

公开(公告)日：2008-02-14

申请号：PCT/US2007/074889

申请日：2007-07-31

Applicant: QUALCOMM INCORPORATED ,  QI, Yingyong

Inventor： QI, Yingyong

IPC: H04N7/50

CPC classification number: H04N19/176 ,  H04N19/54 ,  H04N19/61

Abstract: Techniques for performing mesh-based video compression/decompression with domain transformation are described. A video encoder partitions an image into meshes of pixels, processes the meshes of pixels to obtain blocks of prediction errors, and codes the blocks of prediction errors to generate coded data for the image. The meshes may have arbitrary polygonal shapes and the blocks may have a predetermined shape, e.g., square. The video encoder may process the meshes of pixels to obtain meshes of prediction errors and may then transform the meshes of prediction errors to the blocks of prediction errors. Alternatively, the video encoder may transform the meshes of pixels to blocks of pixels and may then process the blocks of pixels to obtain the blocks of prediction errors. The video encoder may also perform mesh-based motion estimation to determine reference meshes used to generate the prediction errors.

Abstract translation: 描述了使用域转换执行基于网格的视频压缩/解压缩的技术。 视频编码器将图像划分为像素网格，处理像素的网格以获得预测误差块，并且编码预测误差块以生成图像的编码数据。 网格可以具有任意的多边形形状，并且块可以具有预定的形状，例如正方形。 视频编码器可以处理像素的网格以获得预测误差的网格，然后可以将预测误差的网格转换为预测误差块。 或者，视频编码器可以将像素的网格变换为像素块，然后可以处理像素块以获得预测误差块。 视频编码器还可以执行基于网格的运动估计，以确定用于生成预测误差的参考网格。

公开(公告)号：WO2015195301A1

公开(公告)日：2015-12-23

申请号：PCT/US2015/033368

申请日：2015-05-29

Applicant: QUALCOMM INCORPORATED

Inventor： MOHAMED, Magdi Abuelgasim ,  SARKIS, Michel Adib ,  QI, Yingyong ,  ZHONG, Xin ,  BI, Ning

IPC: G06K9/00 ,  G06K9/46 ,  G06T11/20

CPC classification number: G06K9/6212 ,  G06K9/0061 ,  G06K9/4633 ,  G06K9/4642 ,  G06T3/0012 ,  G06T2207/20021

Abstract: A method (500) for obtaining structural information from a digital image by an electronic device is described. The method includes determining (510) an iris position in a region of interest based on a gradient direction transform (504, 506, 508). Determining the iris position may include determining a first dimension position and a second dimension position corresponding to a maximum value in the transform space.

Abstract translation: 描述了一种用于通过电子设备从数字图像获得结构信息的方法（500）。 该方法包括基于梯度方向变换（504,506,508）确定（510）感兴趣区域中的虹膜位置。 确定虹膜位置可以包括确定对应于变换空间中的最大值的第一维度位置和第二维度位置。