-
公开(公告)号：US08155399B2
公开(公告)日：2012-04-10
申请号：US12056051
申请日：2008-03-26
Applicant: Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
Inventor： Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
IPC: G06K9/62
CPC classification number: G06K9/00241 , G06K9/621
Abstract: There is provided a discriminative framework for image alignment. Image alignment is generally the process of moving and deforming a template to minimize the distance between the template and an image. There are essentially three elements to image alignment, namely template representation, distance metric, and optimization method. For template representation, given a face dataset with ground truth landmarks, a boosting-based classifier is trained that is able to learn the decision boundary between two classes—the warped images from ground truth landmarks (e.g., positive class) and those from perturbed landmarks (e.g., negative class). A set of trained weak classifiers based on Haar-like rectangular features determines a boosted appearance model. A distance metric is a score from the strong classifier, and image alignment is the process of optimizing (e.g., maximizing) the classification score. On the generic face alignment problem, the proposed framework greatly improves the robustness, accuracy, and efficiency of alignment.
Abstract translation: 提供了一种用于图像对齐的辨别框架。 图像对齐通常是移动和变形模板的过程，以最小化模板和图像之间的距离。 图像对齐基本上有三个要素，即模板表示，距离度量和优化方法。 对于模板表示，给定一个具有地面真实地标的面部数据集，训练有素的分类器能够学习两个类之间的决策边界 - 来自地面真实地标（例如，积极的类）和来自扰动地标的变形图像 （例如负面班）。 基于哈尔式矩形特征的一组经过训练的弱分类器决定了外观模型的提升。 距离度量是来自强分类器的分数，图像对准是优化（例如，最大化）分类分数的过程。 在通用面对齐问题上，提出的框架大大提高了对齐的鲁棒性，准确性和效率。
-
公开(公告)号：US08498454B2
公开(公告)日：2013-07-30
申请号：US12627039
申请日：2009-11-30
Applicant: Jilin Tu , Frederick Wilson Wheeler , Peter Henry Tu , Xiaoming Liu , Yan Tong
Inventor： Jilin Tu , Frederick Wilson Wheeler , Peter Henry Tu , Xiaoming Liu , Yan Tong
IPC: G06K9/00
CPC classification number: G06K9/6234 , G06K9/00288 , G06K9/6215
Abstract: A technique for optimizing object recognition is disclosed. The technique includes receiving at least one image of an object and at least one reference image. The technique further includes identifying at least one performance metric corresponding to an object recognition task. The identified performance metric is optimized to generate the corresponding optimized performance metric by determining an optimal subspace based on a determined objective function corresponding to the object recognition task and a difference between the received image and the corresponding reference image. Subsequently, the technique includes comparing the received image with the reference image based on the optimized performance metric for performing the object recognition task.
Abstract translation: 公开了一种用于优化对象识别的技术。 该技术包括接收对象和至少一个参考图像的至少一个图像。 该技术还包括识别对应于对象识别任务的至少一个性能量度。 通过基于与对象识别任务相对应的确定的目标函数和接收到的图像与对应的参考图像之间的差异来确定最佳子空间来优化识别的性能度量以产生相应的优化性能度量。 随后，该技术包括基于用于执行对象识别任务的优化性能度量来比较接收到的图像与参考图像。
-
公开(公告)号：US20120237117A1
公开(公告)日：2012-09-20
申请号：US13052097
申请日：2011-03-20
Applicant: Xiaoming Liu , Frederick Wilson Wheeler , Peter Henry Tu , Jilin Tu
Inventor： Xiaoming Liu , Frederick Wilson Wheeler , Peter Henry Tu , Jilin Tu
CPC classification number: G06K9/3241 , G06K9/00228
Abstract: A method for image alignment is disclosed. In one embodiment, the method includes acquiring a facial image of a person and using a discriminative face alignment model to fit a generic facial mesh to the facial image to facilitate locating of facial features. The discriminative face alignment model may include a generative shape model component and a discriminative appearance model component. Further, the discriminative appearance model component may have been trained to estimate a score function that minimizes the angle between a gradient direction and a vector pointing toward a ground-truth shape parameter. Additional methods, systems, and articles of manufacture are also disclosed.
Abstract translation: 公开了一种用于图像对准的方法。 在一个实施例中，该方法包括获取人的面部图像并使用区别性面部对准模型来将面部图像拟合为面部图像，以便于面部特征的定位。 鉴别面部对齐模型可以包括生成形状模型组件和鉴别外观模型组件。 此外，鉴别外观模型组件可能已经被训练以估计最小化梯度方向和指向地面真实形状参数的向量之间的角度的得分函数。 还公开了附加的方法，系统和制品。
-
4.发明授权公开(公告)号：US08442330B2
公开(公告)日：2013-05-14
申请号：US12533066
申请日：2009-07-31
Applicant: Yan Tong , Xiaoming Liu , Jilin Tu , Peter Henry Tu , Frederick Wilson Wheeler
Inventor： Yan Tong , Xiaoming Liu , Jilin Tu , Peter Henry Tu , Frederick Wilson Wheeler
CPC classification number: G06K9/66 , G06K9/00228 , G06T7/33 , G06T7/35 , G06T2207/20121 , G06T2207/30201
Abstract: A system and method for estimating a set of landmarks for a large image ensemble employs only a small number of manually labeled images from the ensemble and avoids labor-intensive and error-prone object detection, tracking and alignment learning task limitations associated with manual image labeling techniques. A semi-supervised least squares congealing approach is employed to minimize an objective function defined on both labeled and unlabeled images. A shape model is learned on-line to constrain the landmark configuration. A partitioning strategy allows coarse-to-fine landmark estimation.
Abstract translation: 用于估计大图像集合的一组地标的系统和方法仅使用来自集合的少量手动标记的图像，并且避免与手动图像标签相关联的劳动密集型和易出错的对象检测，跟踪和对准学习任务限制 技术 采用半监督的最小二乘法凝结方法来最小化在标记和未标记图像上定义的目标函数。 形状模型在线学习以约束地标配置。 分区策略允许粗略到精细的地标估计。
-
公开(公告)号：US20090257625A1
公开(公告)日：2009-10-15
申请号：US12100620
申请日：2008-04-10
Applicant: Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
Inventor： Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
IPC: G06K9/00
CPC classification number: G06K9/00288 , G06K9/00261 , G06K9/621
Abstract: A method for face model fitting comprising, receiving a first observed image, receiving a second observed image, and fitting an active appearance model of a third image to the second observed image and the first observed image with an algorithm that includes a first function of a mean-square-error between a warped image of the second observed image and a synthesis of the active appearance model and a second function of a mean-square-error between the warped image of the second observed image and an appearance data of the first observed image.
Abstract translation: 一种用于面部模型拟合的方法，包括：接收第一观察图像，接收第二观察图像，以及使用包括第一观察图像的第一函数的第二观察图像和第一观察图像拟合第三图像的活动外观模型 第二观察图像的翘曲图像与活动外观模型的合成之间的均方误差以及第二观察图像的翘曲图像与第一观察图像的外观数据之间的均方误差的第二函数 图片。
-
Patent Agency Ranking
6.发明申请公开(公告)号：US20080175509A1
公开(公告)日：2008-07-24
申请号：US11836316
申请日：2007-08-09
Applicant: Frederick Wilson Wheeler , Xiaoming Liu , Peter Henry Tu
Inventor： Frederick Wilson Wheeler , Xiaoming Liu , Peter Henry Tu
IPC: G06K9/40
CPC classification number: G06K9/621 , G06K9/00281 , G06T7/33
Abstract: A system and method for performing a facial image restoration is described. The system includes an active appearance model component for fitting an active appearance model to a facial image found in each of a plurality of video frames, a registration component for registering each pixel of each facial image with comparable pixels of each of the other facial images, and a restoration component for producing a restored facial image from the facial images. The method includes fitting an active appearance model to a facial image found in each of a plurality of video frames, registering each pixel of each said facial image with comparable pixels of each of the other facial images, and producing a restored facial image from the facial images.
Abstract translation: 描述用于执行面部图像恢复的系统和方法。 该系统包括用于将活动外观模型拟合到在多个视频帧中的每一个中发现的面部图像的活动外观模型部件，用于将每个面部图像的每个像素与每个其他面部图像的可比较像素对准的注册部件， 以及恢复部件，用于从面部图像产生恢复的面部图像。 该方法包括将活动外观模型拟合到在多个视频帧中的每一个中发现的面部图像，将每个所述面部图像的每个像素与每个其他面部图像的可比较像素对准，并从面部产生恢复的面部图像 图片。
-
公开(公告)号：US08224037B2
公开(公告)日：2012-07-17
申请号：US12100620
申请日：2008-04-10
Applicant: Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
Inventor： Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
IPC: G06K9/00
CPC classification number: G06K9/00288 , G06K9/00261 , G06K9/621
Abstract: A method for face model fitting comprising, receiving a first observed image, receiving a second observed image, and fitting an active appearance model of a third image to the second observed image and the first observed image with an algorithm that includes a first function of a mean-square-error between a warped image of the second observed image and a synthesis of the active appearance model and a second function of a mean-square-error between the warped image of the second observed image and an appearance data of the first observed image.
Abstract translation: 一种用于面部模型拟合的方法，包括：接收第一观察图像，接收第二观察图像，以及使用包括第一观察图像的第一函数的第二观察图像和第一观察图像拟合第三图像的活动外观模型 第二观察图像的翘曲图像与活动外观模型的合成之间的均方误差以及第二观察图像的翘曲图像与第一观察图像的外观数据之间的均方误差的第二函数 图片。
-
8.发明授权公开(公告)号：US08064712B2
公开(公告)日：2011-11-22
申请号：US11836316
申请日：2007-08-09
Applicant: Frederick Wilson Wheeler , Xiaoming Liu , Peter Henry Tu
Inventor： Frederick Wilson Wheeler , Xiaoming Liu , Peter Henry Tu
IPC: G06K9/40
CPC classification number: G06K9/621 , G06K9/00281 , G06T7/33
Abstract: A system and method for performing a facial image restoration is described. The system includes an active appearance model component for fitting an active appearance model to a facial image found in each of a plurality of video frames, a registration component for registering each pixel of each facial image with comparable pixels of each of the other facial images, and a restoration component for producing a restored facial image from the facial images. The method includes fitting an active appearance model to a facial image found in each of a plurality of video frames, registering each pixel of each said facial image with comparable pixels of each of the other facial images, and producing a restored facial image from the facial images.
Abstract translation: 描述用于执行面部图像恢复的系统和方法。 该系统包括用于将活动外观模型拟合到在多个视频帧中的每一个中发现的面部图像的活动外观模型部件，用于将每个面部图像的每个像素与每个其他面部图像的可比较像素对准的注册部件， 以及恢复部件，用于从面部图像产生恢复的面部图像。 该方法包括将活动外观模型拟合到在多个视频帧中的每一个中发现的面部图像，将每个所述面部图像的每个像素与每个其他面部图像的可比较像素对准，并从面部产生恢复的面部图像 图片。
-
9.发明授权Method of combining images of multiple resolutions to produce an enhanced active appearance model 有权组合多个分辨率图像以产生增强的活动外观模型的方法公开(公告)号：US07885455B2
公开(公告)日：2011-02-08
申请号：US11650213
申请日：2007-01-05
Applicant: Xiaoming Liu , Frederick Wilson Wheeler , Peter Henry Tu
Inventor： Xiaoming Liu , Frederick Wilson Wheeler , Peter Henry Tu
CPC classification number: G06K9/621 , G06K9/00281
Abstract: A method of producing an enhanced Active Appearance Model (AAM) by combining images of multiple resolutions is described herein. The method generally includes processing a plurality of images each having image landmarks and each image having an original resolution level. The images are down-sampled into multiple scales of reduced resolution levels. The AAM is trained for each image at each reduced resolution level, thereby creating a multi-resolution AAM. An enhancement technique is then used to refine the image landmarks for training the AAM at the original resolution level. The landmarks for training the AAM at each level of reduced resolution is obtained by scaling the landmarks used at the original resolution level by a ratio in accordance with the multiple scales.
Abstract translation: 本文描述了通过组合多个分辨率的图像来生成增强的活动外观模型（AAM）的方法。 该方法通常包括处理多个具有图像界标的图像，并且每个图像具有原始分辨率级别。 图像被下采样成分辨率降低的多个尺度。 在每个降低的分辨率级别对每个图像训练AAM，从而创建多分辨率AAM。 然后使用增强技术来改善用于以原始分辨率级别训练AAM的图像界标。 通过将原始分辨率级别使用的地标按照多个尺度的比例进行缩放，可以获得在每个降低分辨率级别下对AAM进行训练的地标。
-
公开(公告)号：US20080310759A1
公开(公告)日：2008-12-18
申请号：US12056051
申请日：2008-03-26
Applicant: Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
Inventor： Xiaoming Liu , Peter Henry Tu , Frederick Wilson Wheeler
IPC: G06K9/32
CPC classification number: G06K9/00241 , G06K9/621
Abstract: There is provided a discriminative framework for image alignment. Image alignment is generally the process of moving and deforming a template to minimize the distance between the template and an image. There are essentially three elements to image alignment, namely template representation, distance metric, and optimization method. For template representation, given a face dataset with ground truth landmarks, a boosting-based classifier is trained that is able to learn the decision boundary between two classes-the warped images from ground truth landmarks (e.g., positive class) and those from perturbed landmarks (e.g., negative class). A set of trained weak classifiers based on Haar-like rectangular features determines a boosted appearance model. A distance metric is a score from the strong classifier, and image alignment is the process of optimizing (e.g., maximizing) the classification score. On the generic face alignment problem, the proposed framework greatly improves the robustness, accuracy, and efficiency of alignment.
Abstract translation: 提供了一种用于图像对齐的辨别框架。 图像对齐通常是移动和变形模板的过程，以最小化模板和图像之间的距离。 图像对齐基本上有三个要素，即模板表示，距离度量和优化方法。 对于模板表示，给定一个具有地面真实地标的面部数据集，训练有素的分类器能够学习两个类之间的决策边界 - 来自地面真实地标（例如，积极的类）和来自扰动地标的变形图像 （例如负面班）。 基于哈尔式矩形特征的一组经过训练的弱分类器决定了外观模型的提升。 距离度量是来自强分类器的分数，图像对准是优化（例如，最大化）分类分数的过程。 在通用面对齐问题上，提出的框架大大提高了对齐的鲁棒性，准确性和效率。
-
-
-
-
-
-
-
-
-
Search Results
59
records
(took 0.026 seconds)
