公开(公告)号：US08244044B2

公开(公告)日：2012-08-14

申请号：US12109347

申请日：2008-04-25

申请人： Gang Hua ,  Paul Viola ,  David Liu

发明人： Gang Hua ,  Paul Viola ,  David Liu

IPC分类号： G06K9/62 ,  G06K9/46

CPC分类号： G06K9/4647 ,  G06K9/468 ,  G06K9/6228

摘要： Image feature selection and extraction (e.g., for image classifier training) is accomplished in an integrated manner, such that higher-order features are merely developed from first-order features selected for image classification. That is, first-order image features are selected for image classification from an image feature pool, initially populated with pre-extracted first-order image features. The selected first-order classifying features are paired with previously selected first-order classifying features to generate higher-order features. The higher-order features are placed into the image feature pool as they are developed or “on-the-fly” (e.g., for use in image classifier training).

摘要翻译： 图像特征选择和提取（例如，用于图像分类器训练）以集成的方式实现，使得仅从为图像分类选择的一阶特征开发高阶特征。 也就是说，从图像特征池中选择用于图像分类的一阶图像特征，最初用预提取的一阶图像特征填充。 所选择的一阶分类特征与先前选择的一阶分类特征配对以产生更高阶的特征。 更高阶的特征被放置在图像特征池中，因为它们被开发或“即时”（例如，用于图像分类器训练）。

公开(公告)号：US20090316986A1

公开(公告)日：2009-12-24

申请号：US12109347

申请日：2008-04-25

申请人： Gang Hua ,  Paul Viola ,  David Liu

发明人： Gang Hua ,  Paul Viola ,  David Liu

IPC分类号： G06K9/46 ,  G06K9/62

CPC分类号： G06K9/4647 ,  G06K9/468 ,  G06K9/6228

摘要： Image feature selection and extraction (e.g., for image classifier training) is accomplished in an integrated manner, such that higher-order features are merely developed from first-order features selected for image classification. That is, first-order image features are selected for image classification from an image feature pool, initially populated with pre-extracted first-order image features. The selected first-order classifying features are paired with previously selected first-order classifying features to generate higher-order features. The higher-order features are placed into the image feature pool as they are developed or “on-the-fly” (e.g., for use in image classifier training).

摘要翻译： 图像特征选择和提取（例如，用于图像分类器训练）以集成的方式实现，使得仅从为图像分类选择的一阶特征开发高阶特征。 也就是说，从图像特征池中选择用于图像分类的一阶图像特征，最初用预提取的一阶图像特征填充。 所选择的一阶分类特征与先前选择的一阶分类特征配对以产生更高阶的特征。 更高阶的特征被放置在图像特征池中，因为它们被开发或“即时”（例如，用于图像分类器训练）。

公开(公告)号：US08891861B2

公开(公告)日：2014-11-18

申请号：US13371719

申请日：2012-02-13

申请人： Gang Hua ,  Paul Viola

发明人： Gang Hua ,  Paul Viola

IPC分类号： G06K9/62 ,  G06F17/30 ,  G06K9/00 ,  G06K9/46

CPC分类号： G06F17/3025 ,  G06F17/30262 ,  G06K9/00664 ,  G06K9/4642 ,  G06K9/4652 ,  G06K9/6256 ,  G06K9/6285

摘要： Images are classified as photos (e.g., natural photographs) or graphics (e.g., cartoons, synthetically generated images), such that when searched (online) with a filter, an image database returns images corresponding to the filter criteria (e.g., either photos or graphics will be returned). A set of image statistics pertaining to various visual cues (e.g., color, texture, shape) are identified in classifying the images. These image statistics, combined with pre-tagged image metadata defining an image as either a graphic or a photo, may be used to train a boosting decision tree. The trained boosting decision tree may be used to classify additional images as graphics or photos based on image statistics determined for the additional images.

摘要翻译： 图像被分类为照片（例如，自然照片）或图形（例如，漫画，综合生成的图像）​​，使得当用过滤器搜索（在线）时，图像数据库返回与过滤标准相对应的图像（例如，照片或 图形将被返回）。 在对图像进行分类时，识别关于各种视觉提示（例如，颜色，纹理，形状）的一组图像统计信息。 这些图像统计信息与将图像定义为图形或照片的预先标记的图像元数据可以用于训练增强决策树。 经训练的增强决策树可以用于基于为附加图像确定的图像统计来将附加图像分类为图形或照片。

公开(公告)号：US08131066B2

公开(公告)日：2012-03-06

申请号：US12098026

申请日：2008-04-04

申请人： Gang Hua ,  Paul Viola

发明人： Gang Hua ,  Paul Viola

IPC分类号： G06K9/62

CPC分类号： G06F17/3025 ,  G06F17/30262 ,  G06K9/00664 ,  G06K9/4642 ,  G06K9/4652 ,  G06K9/6256 ,  G06K9/6285

摘要： Images are classified as photos (e.g., natural photographs) or graphics (e.g., cartoons, synthetically generated images), such that when searched (online) with a filter, an image database returns images corresponding to the filter criteria (e.g., either photos or graphics will be returned). A set of image statistics pertaining to various visual cues (e.g., color, texture, shape) are identified in classifying the images. These image statistics, combined with pre-tagged image metadata defining an image as either a graphic or a photo, may be used to train a boosting decision tree. The trained boosting decision tree may be used to classify additional images as graphics or photos based on image statistics determined for the additional images.

摘要翻译： 图像被分类为照片（例如，自然照片）或图形（例如，漫画，综合生成的图像）​​，使得当用过滤器搜索（在线）时，图像数据库返回与过滤标准相对应的图像（例如，照片或 图形将被返回）。 在对图像进行分类时，识别关于各种视觉提示（例如，颜色，纹理，形状）的一组图像统计信息。 这些图像统计信息与将图像定义为图形或照片的预先标记的图像元数据可以用于训练增强决策树。 经训练的增强决策树可以用于基于为附加图像确定的图像统计来将附加图像分类为图形或照片。

公开(公告)号：US20120141020A1

公开(公告)日：2012-06-07

申请号：US13371719

申请日：2012-02-13

申请人： Gang Hua ,  Paul Viola

发明人： Gang Hua ,  Paul Viola

IPC分类号： G06K9/62

CPC分类号： G06F17/3025 ,  G06F17/30262 ,  G06K9/00664 ,  G06K9/4642 ,  G06K9/4652 ,  G06K9/6256 ,  G06K9/6285

摘要： Images are classified as photos (e.g., natural photographs) or graphics (e.g., cartoons, synthetically generated images), such that when searched (online) with a filter, an image database returns images corresponding to the filter criteria (e.g., either photos or graphics will be returned). A set of image statistics pertaining to various visual cues (e.g., color, texture, shape) are identified in classifying the images. These image statistics, combined with pre-tagged image metadata defining an image as either a graphic or a photo, may be used to train a boosting decision tree. The trained boosting decision tree may be used to classify additional images as graphics or photos based on image statistics determined for the additional images.

摘要翻译： 图像被分类为照片（例如，自然照片）或图形（例如，漫画，综合生成的图像）​​，使得当用过滤器搜索（在线）时，图像数据库返回与过滤标准相对应的图像（例如，照片或 图形将被返回）。 在对图像进行分类时，识别关于各种视觉提示（例如，颜色，纹理，形状）的一组图像统计信息。 这些图像统计信息与将图像定义为图形或照片的预先标记的图像元数据可以用于训练增强决策树。 经训练的增强决策树可以用于基于为附加图像确定的图像统计来将附加图像分类为图形或照片。

公开(公告)号：US20090252413A1

公开(公告)日：2009-10-08

申请号：US12098026

申请日：2008-04-04

申请人： Gang Hua ,  Paul Viola

发明人： Gang Hua ,  Paul Viola

IPC分类号： G06K9/00 ,  G06K9/62

CPC分类号： G06F17/3025 ,  G06F17/30262 ,  G06K9/00664 ,  G06K9/4642 ,  G06K9/4652 ,  G06K9/6256 ,  G06K9/6285

摘要： Images are classified as photos (e.g., natural photographs) or graphics (e.g., cartoons, synthetically generated images), such that when searched (online) with a filter, an image database returns images corresponding to the filter criteria (e.g., either photos or graphics will be returned). A set of image statistics pertaining to various visual cues (e.g., color, texture, shape) are identified in classifying the images. These image statistics, combined with pre-tagged image metadata defining an image as either a graphic or a photo, may be used to train a boosting decision tree. The trained boosting decision tree may be used to classify additional images as graphics or photos based on image statistics determined for the additional images.

摘要翻译： 图像被分类为照片（例如，自然照片）或图形（例如，漫画，综合生成的图像）​​，使得当用过滤器搜索（在线）时，图像数据库返回与过滤标准相对应的图像（例如，照片或 图形将被返回）。 在对图像进行分类时，识别关于各种视觉提示（例如，颜色，纹理，形状）的一组图像统计信息。 这些图像统计信息与将图像定义为图形或照片的预先标记的图像元数据可以用于训练增强决策树。 经训练的增强决策树可以用于基于为附加图像确定的图像统计来将附加图像分类为图形或照片。

公开(公告)号：US08644563B2

公开(公告)日：2014-02-04

申请号：US12637494

申请日：2009-12-14

申请人： Amir Akbarzadeh ,  Gang Hua

发明人： Amir Akbarzadeh ,  Gang Hua

IPC分类号： G06K9/00

CPC分类号： G06K9/00288 ,  G06K9/00677 ,  G06K9/6212

摘要： Face recognition may be performed using a combination of visual analysis and social context. In one example, a web site such as a social networking site or photo-sharing site allows users to upload photos, and allows faces that appear in the photo to be tagged with users' names. When user A uploads a new photo, two analyses may be performed. First, a face in the photo is compared with known faces of users to determine similarity. Second, it is determined which other users user A frequently uploads photos of. Two probability distributions are created. One distribution assigns high probabilities to users whose photos are similar to the new photo. The other assigns high probabilities to users who frequently appear in photos uploaded by user A. These probability distributions are combined, and the person in the photo is identified as being the person with the highest probability.

摘要翻译： 可以使用视觉分析和社会语境的组合来执行面部识别。 在一个示例中，诸如社交网站或照片共享网站的网站允许用户上传照片，并且允许照片中出现的脸部被用户的姓名标记。 当用户A上传新照片时，可能会执行两次分析。 首先，将照片中的脸部与用户的已知脸部进行比较，以确定相似性。 其次，确定哪些其他用户A经常上传照片。 创建两个概率分布。 一个分配对于照片类似于新照片的用户分配高概率。 另一方则将频繁出现在用户A上传的照片中的用户分配给高概率。这些概率分布相结合，照片中的人被识别为具有最高概率的人。

公开(公告)号：US20120159404A1

公开(公告)日：2012-06-21

申请号：US13398645

申请日：2012-02-16

申请人： Srinath Vasireddy ,  Sergey Chub ,  Gang Hua ,  Ting-yi Yang

发明人： Srinath Vasireddy ,  Sergey Chub ,  Gang Hua ,  Ting-yi Yang

IPC分类号： G06F3/033

CPC分类号： H04N19/533 ,  G06T7/223 ,  G06T2207/10016 ,  H04N19/56

摘要： A processing device and method are provided for capturing images, via an image-capturing component of a processing device, and determining a motion of the processing device. An adaptive search center technique may be employed to determine a search center with respect to multiple equal-sized regions of an image frame, based on previously estimated motion vectors. One of several fast block matching methods may be used, based on one or more conditions, to match a block of pixels of one image frame with a second block of pixels of a second image. Upon matching blocks of pixels, motion vectors of the multiple equal-sized regions may be estimated. The motion may be determined, based on the estimated motion vectors, and an associated action may be performed. Various embodiments may implement techniques to distinguish motion blur from de-focus blur and to determine a change in lighting condition.

摘要翻译： 提供了一种处理装置和方法，用于经由处理装置的图像捕获部件捕获图像，并且确定处理装置的运动。 可以采用自适应搜索中心技术来基于先前估计的运动矢量来确定关于图像帧的多个等大小区域的搜索中心。 可以使用几种快速块匹配方法中的一种，基于一个或多个条件来匹配一个图像帧的像素块与第二图像的第二像素块。 在匹配像素块之后，可以估计多个等大小区域的运动矢量。 可以基于估计的运动矢量来确定运动，并且可以执行相关联的动作。 各种实施例可以实现将运动模糊与去焦点模糊区分开来并且确定照明条件的变化的技术。

公开(公告)号：US07936906B2

公开(公告)日：2011-05-03

申请号：US11763909

申请日：2007-06-15

申请人： Gang Hua ,  Paul A Viola ,  Steven M. Drucker ,  Michael Revow

发明人： Gang Hua ,  Paul A Viola ,  Steven M. Drucker ,  Michael Revow

IPC分类号： G06K9/00

CPC分类号： G06K9/00288 ,  G06K9/6232

摘要： Systems and methods are described for face recognition using discriminatively trained orthogonal rank one tensor projections. In an exemplary system, images are treated as tensors, rather than as conventional vectors of pixels. During runtime, the system designs visual features—embodied as tensor projections—that minimize intraclass differences between instances of the same face while maximizing interclass differences between the face and faces of different people. Tensor projections are pursued sequentially over a training set of images and take the form of a rank one tensor, i.e., the outer product of a set of vectors. An exemplary technique ensures that the tensor projections are orthogonal to one another, thereby increasing ability to generalize and discriminate image features over conventional techniques. Orthogonality among tensor projections is maintained by iteratively solving an ortho-constrained eigenvalue problem in one dimension of a tensor while solving unconstrained eigenvalue problems in additional dimensions of the tensor.

摘要翻译： 使用区分训练的正交秩一张量投影描述用于人脸识别的系统和方法。 在示例性系统中，图像被视为张量，而不是像传统的像素矢量。 在运行期间，系统设计视觉特征 - 体现为张量投影 - 最大限度地减少不同人脸部和脸部之间的类间差异，从而最大限度地减少同一脸部实例之间的差异。 张量投影在训练图像集上顺序追溯，并采取一级张量的形式，即一组向量的外积。 示例性技术确保张量投影彼此正交，从而增加了与常规技术相比的概括和区分图像特征的能力。 通过迭代求解张量的一维中的邻域约束特征值问题，同时解决张量的附加维度中的无约束特征值问题，维持张量投影中的正交性。

公开(公告)号：US20100246969A1

公开(公告)日：2010-09-30

申请号：US12410469

申请日：2009-03-25

申请人： Simon A. J. Winder ,  Gang Hua

发明人： Simon A. J. Winder ,  Gang Hua

IPC分类号： G06K9/46

CPC分类号： G06K9/4671

摘要： Described is a technology in which an image (or image patch) is processed into a highly discriminative and computationally efficient image descriptor that has a low storage footprint. Feature vectors are generated from an image (or image patch), and further processed via a polar Gaussian pooling approach (a DAISY configuration) into a descriptor. The descriptor is normalized, and processed with a dimension reduction component and a quantization component (based upon dynamic range reduction) into a finalized descriptor, which may be further compressed. The resulting descriptors have significantly reduced error rates and significantly smaller sizes than other image descriptors (such as SIFT-based descriptors).

摘要翻译： 描述了一种技术，其中图像（或图像贴片）被处理成具有低存储空间的高度辨别性和计算效率的图像描述符。 特征向量从图像（或图像块）生成，并且通过极高斯混合方法（DAISY配置）进一步处理成描述符。 描述符被归一化，并且将尺寸减小分量和量化分量（基于动态范围缩小）处理成最终描述符，其可被进一步压缩。 所得到的描述符与其他图像描述符（例如基于SIFT的描述符）相比，显着降低了错误率和显着更小的大小。