公开(公告)号：EP2960860A1

公开(公告)日：2015-12-30

申请号：EP14305994.7

申请日：2014-06-24

申请人： Thomson Licensing

发明人： Marquant, Gwenaëlle ,  Hellier, Pierre ,  Oisel, Lionel

IPC分类号： G06T7/00 ,  G06T11/60

CPC分类号： G06T11/60 ,  G06T7/12 ,  G06T7/155 ,  G06T2207/20096 ,  G06T2207/20116 ,  G06T2207/20132

摘要： A method and an apparatus for cloning an object from a source image to a target image are suggested. The suggested method comprises: creating a segmentation curve for the object to be cloned in the source image; calculating a stopping criterion as a function of the spatial gradients in the surroundings of the segmentation curve; adapting the segmentation curve according to a comparison between the stopping criterion and a threshold; and cloning the object with the adapted segmentation curve to the target image. According to the suggested method and an apparatus, a user can be provided with at least one modified segmentation curve adapted according to a comparison between a stopping criterion as a function of the spatial gradients in the surroundings of the segmentation curve and a threshold. Upon the validation of the user, the modified segmentation curve is used for the cloning process, which can improve the subjective quality of the cloning even with low-quality segmentation.

摘要翻译： 提出了一种用于将对象从源图像克隆到目标图像的方法和装置。 所提出的方法包括：为源图像中要克隆的对象创建分割曲线; 计算作为分割曲线周围空间梯度的函数的停止标准; 根据停止准则和阈值之间的比较来适应分割曲线; 并将具有适应的分割曲线的对象克隆到目标图像。 根据所提出的方法和装置，可以向用户提供至少一个修改的分割曲线，该曲线根据分割曲线的周围空间梯度的停止准则与阈值之间的比较来适应。 在验证用户后，修改的分割曲线用于克隆过程，即使使用低质量分割，也可以提高克隆的主观质量。

公开(公告)号：EP2890109A1

公开(公告)日：2015-07-01

申请号：EP14200011.6

申请日：2014-12-23

申请人： Thomson Licensing

发明人： Hellier, Pierre ,  Sabater, Neus ,  Demoulin, Vincent

IPC分类号： H04N1/60 ,  G06F17/30

CPC分类号： G06F17/3025 ,  G06F17/30256 ,  G06T5/00 ,  G06T7/90 ,  H04N1/60 ,  H04N1/6063

摘要： The invention relates to a method for sorting a group of images of a database, the images of said group being called example images and being associated to a same scene.
According to the invention, said method comprising determining and assigning a score to each example image and sorting the group of images according to the assigned scores, the method is characterized in that determining a score to each example image comprises:
- obtaining (4112) an information ( C D , E, E i , D color i ) representing at least a color distribution of at least one part of the source image; and
- for each example image:
▪ obtaining (4111) a value ( V sci , N sci ) representing a spatial correspondence between the example image and the source image ( I S ) representing at least a part of the scene; and
▪ determining (411) a score ( Score i ) for the example image from said value and said information.

摘要翻译： 本发明涉及一种用于分选的基团的数据库的图像的方法，所述组的图像被称为实施例的图像和被关联于同一场景。 。根据本发明，所述方法包括确定开采和将一个得分指定给每个实例图像和分选图像组雅丁到分配的分数，该方法在没有确定性采矿的得分给每个示例图像包括为特征： - 获得（4112），以 信息（CD，E，E I，D颜色ⅰ）表示至少所述源图像中的至少一个部分的颜色分布; 和 - 对于每个实施例的图像：-ª获取（4111）的值（V SCI，SCI N）表示所述实例图像和表示至少所述场景的一部分的源图像（I S）之间的空间对应关系; 和-ª确定性开采（411）的得分（分数i）用于从所述值和信息的示例性图像。

公开(公告)号：EP2819095A1

公开(公告)日：2014-12-31

申请号：EP13305859.4

申请日：2013-06-24

申请人： Thomson Licensing

发明人： Hellier, Pierre ,  Chamaret, Christel ,  Oisel, Lionel

IPC分类号： G06T11/60 ,  H04N5/272

CPC分类号： H04N21/812 ,  G06T11/60 ,  H04N5/2723 ,  H04N21/845

摘要： A method and an apparatus for inserting a virtual object in a video are described. The method utilizes a saliency map that characterizes the gaze allocation from a viewer on the video and inserts the virtual object in the video based on the saliency map. The method comprises the steps of: generating (12) a saliency map of the video after the insertion (11) of the virtual object; and adjusting (13) the insertion of the virtual object based on the saliency map by adjusting at least one visual characteristic of the inserted virtual object.

摘要翻译： 描述用于在视频中插入虚拟对象的方法和装置。 该方法利用显着性映射来表征视频上的观看者的注视分配，并根据显着性图将虚拟对象插入到视频中。 该方法包括以下步骤：在虚拟对象的插入（11）之后生成（12）视频的显着图; 以及通过调整所插入的虚拟对象的至少一个视觉特征来调整（13）基于所述显着性图来插入虚拟对象。

公开(公告)号：EP3021282A1

公开(公告)日：2016-05-18

申请号：EP14306813.8

申请日：2014-11-14

申请人： Thomson Licensing

发明人： Hellier, Pierre ,  Sabater, Neus ,  Perez, Patrick

IPC分类号： G06T11/00

CPC分类号： G06T11/001

摘要： A particular implementation determines color palettes of images by extracting and decomposing color palettes based on the image color content. The decomposition can produce a dictionary matrix, an activation matrix, or both. The dictionary matrix can be used in recoloring an image, either directly or after storing. Another implementation selects a color palette to recolor an image by accessing metadata associated with the image and estimating a scene type based on the metadata and/or other information. Color palettes are retrieved from memory corresponding to the scene type of the image and are used for recoloring the image. Instructions for the implementation can be stored on a non-transitory computer readable medium such that the embodiments can be implemented by one or more processors.

摘要翻译： 特定实现bestimmt图像的调色板通过提取并基于所述彩色图像内容分解调色板。 分解可产生字典矩阵活化基质，或两者。 字典基质可以在图像重着色中使用，无论是直接或存放之后。 另一种实现中，选择一个调色板通过访问与图像相关联的元数据以及基于所述元数据和/或其它信息的场景类型重新着色的图像。 调色板从存储器对应于所述场景类型的图像的检索并用于重新着色的图像。 指令的执行可被存储并检查该实施例可以由一个或多个处理器中实现的非暂时性计算机可读介质上。

公开(公告)号：EP2953343A1

公开(公告)日：2015-12-09

申请号：EP14305837.8

申请日：2014-06-03

申请人： Thomson Licensing

发明人： Sabater, Neus ,  Hellier, Pierre ,  Demoulin, Vincent

IPC分类号： H04N1/62 ,  G06F17/30

CPC分类号： G06F17/30256 ,  H04N1/60 ,  H04N1/62

摘要： The invention relates to a method for homogenizing at least one image feature of a group of N images, N being an integer such that N > 2,
According to the invention, said method implements the following steps:
- clustering (11) said group of N images delivering at least one cluster to homogenize comprising at least three images of said group of N images, and presenting a distance between said at least three images lower than a maximal predetermined value D,
- from said at least one cluster to homogenize, determining (12) an example image,
- using said example image for homogenizing (13) said at least one image feature in said at least three images of said at least one cluster to homogenize.

摘要翻译： 本发明涉及一种用于均匀化的一组N个图像中的至少一个图像特征，N是整数搜索做N> 2。根据本发明，所述方法实现了以下步骤： - 群集（11）所述的组 N个图像递送至少一个集群以匀化包含所述组N个图像中的至少三幅图像，并呈现比最大预定值D时至少三幅图像之间的所述距离， - 从所述至少一个集群以均质化的，确定性的挖掘 （12）在例如图像 - 使用所述示例图像均匀化（13）所述至少一个图像特征在所述的至少三幅图像的至少一个集群以匀化。

公开(公告)号：EP2819096A1

公开(公告)日：2014-12-31

申请号：EP14172905.3

申请日：2014-06-18

申请人： Thomson Licensing

发明人： Hellier, Pierre ,  Chamaret, Christel ,  Oisel, Lionel

IPC分类号： G06T11/60 ,  H04N5/272

CPC分类号： H04N21/812 ,  G06T11/60 ,  H04N5/2723 ,  H04N21/845

摘要： A method and an apparatus for inserting a virtual object in a video are described. The method utilizes a saliency map that characterizes the gaze allocation from a viewer on an image of the video and inserts the virtual object in the image of the video based on the saliency map. The method comprises: generating (12) a saliency map of the image of the video after the insertion (11) of the virtual object; and adjusting (13) the insertion of the virtual object based on the saliency map by adjusting at least one visual characteristic of the inserted virtual object.

摘要翻译： 描述了用于在视频中插入虚拟对象的方法和设备。 该方法利用表征来自视频的图像上的观看者的注视分配的显着图，并且基于显着图将虚拟对象插入到视频的图像中。 该方法包括：在插入虚拟对象（11）之后生成（12）视频的图像的显着图; 以及通过调整所插入的虚拟对象的至少一个视觉特征来基于所述显着图来调整（13）所述虚拟对象的插入。

公开(公告)号：EP3046318A1

公开(公告)日：2016-07-20

申请号：EP15305033.1

申请日：2015-01-15

申请人： Thomson Licensing

发明人： Hellier, Pierre ,  Frigo, Oriel ,  Sabater, Neus ,  Delon, Julie

IPC分类号： H04N5/235 ,  H04N5/243 ,  H04N9/73

CPC分类号： H04N9/73 ,  H04N5/145 ,  H04N9/67

摘要： One general aspect for motion-based video tonal stabilization uses a keyframe and motion estimation techniques to determine the level of spatial correspondence between input images and the keyframe. When the level of spatial correspondence is high, tonal stabilization is performed through regression and power law tonal transformation to minimize the color differences between images caused by automatic camera parameters without a priori knowledge of the camera model. Tonal error accumulation is reduced by using long-term tonal propagation.

摘要翻译： 基于运动的视频音调稳定的一个一般方面使用关键帧和运动估计技术来确定输入图像和关键帧之间的空间对应关系的水平。 当空间对应关系很高时，通过回归和幂律色调变换进行色调稳定，以最小化由自动摄像机参数引起的图像之间的色差，而无需相机模型的先验知识。 通过使用长期色调传播来降低色调误差累积。

公开(公告)号：EP3035286A1

公开(公告)日：2016-06-22

申请号：EP14307087.8

申请日：2014-12-18

申请人： Thomson Licensing

发明人： Marquant, Gwenaelle ,  Oisel, Lionel ,  Hellier, Pierre

IPC分类号： G06T7/00 ,  G06F3/0484 ,  G06F3/0488

CPC分类号： G06T7/12 ,  G06F3/04845 ,  G06F3/04883 ,  G06T2207/20104

摘要： A method, performed by a touch screen device, for delineating an object in an image is disclosed. The method comprises obtaining, from a user input, a band around the object wherein at least one characteristic of the band depends on the presence of artifacts in the neighborhood of the object to delineate.
A device for implementing the method and a computer program product are further disclosed.

摘要翻译： 公开了一种由触摸屏设备执行的用于描绘图像中的对象的方法。 所述方法包括从用户输入获得所述对象周围的频带，其中所述频带的至少一个特征取决于要描绘的对象附近的伪影的存在。 进一步公开了一种用于实现该方法和计算机程序产品的装置。

公开(公告)号：EP3032816A1

公开(公告)日：2016-06-15

申请号：EP14307011.8

申请日：2014-12-12

申请人： Thomson Licensing

发明人： Briand, Gerard ,  Hellier, Pierre ,  Mollet, Nicolas

IPC分类号： H04N1/60

CPC分类号： H04N1/6088

摘要： The present disclosure relates to a method for displaying at least one image of a video representing a scene, said video being derived from a first video representing said scene acquired at a first date under first lighting conditions, the method comprising:
- acquiring (S1) at least one image, called reference image, of said scene at a second date under second lighting conditions, said second date being subsequent to said first date,
- determining (S2) a color transformation matrix M to be applied to the color values of pixels of at least one image of the first video,
- applying (S3) said color transformation matrix to said at least one image of the first video in order to make the lighting conditions of the scene in the transformed image correspond substantially to the second lighting conditions, and
- displaying (S4) the transformed image.

摘要翻译： 本公开涉及一种用于显示表示场景的视频的至少一个图像的方法，所述视频是从表示在第一照明条件下的第一日期获取的所述场景的第一视频导出的，所述方法包括： - 获取（S1） 在第二照明条件下的第二日期，将所述场景的至少一个图像称为参考图像，所述第二日期在所述第一日期之后， - 确定（S2）应用于像素的颜色值的颜色变换矩阵M 的第一视频的至少一个图像， - 将（S3）所述颜色变换矩阵应用于所述第一视频的所述至少一个图像，以使得所述变换图像中的场景的照明条件基本上对应于所述第二照明条件 ，和 - 显示（S4）变换图像。

公开(公告)号：EP2784736A1

公开(公告)日：2014-10-01

申请号：EP13305387.6

申请日：2013-03-28

申请人： Thomson Licensing

发明人： Hellier, Pierre ,  Démoulin, Vincent ,  Oisel, Lionel ,  Pérez, Patrick

IPC分类号： G06Q30/02 ,  G06F17/30

CPC分类号： G06F17/30247 ,  G06Q30/02

摘要： A method of providing information about an object to a user includes capturing a live image of a scene including the object and reproducing the live image on a screen. A single user input targeting the object on the screen is received, and a still image is captured in response to the user input. A single object is identified in the image in accordance with the location of the single user input, and information about the previously identified object is retrieved from a database. The information related to the object retrieved from the database is reproduced to the user.

摘要翻译： 向用户提供关于对象的信息的方法包括捕获包括对象的场景的实时图像并在屏幕上再现实况图像。 接收到针对屏幕上的对象的单个用户输入，并响应于用户输入捕获静止图像。 根据单个用户输入的位置在图像中识别单个对象，并且从数据库检索关于先前识别的对象的信息。 与从数据库检索到的对象相关的信息被再现给用户。