公开(公告)号：US08873812B2

公开(公告)日：2014-10-28

申请号：US13567309

申请日：2012-08-06

申请人： Diane Larlus-Larrondo ,  Rui Hu ,  Craig Saunders ,  Gabriela Csurka

发明人： Diane Larlus-Larrondo ,  Rui Hu ,  Craig Saunders ,  Gabriela Csurka

IPC分类号： G06K9/00

CPC分类号： G06K9/00 ,  G06T7/11 ,  G06T7/162 ,  G06T7/194 ,  G06T2207/20076

摘要： An image segmentation method includes generating a hierarchy of regions by unsupervised segmentation of an input image. Each region is described with a respective region feature vector representative of the region. Hierarchical structures are identified, each including a parent region and its respective child regions in the hierarchy. Each hierarchical structure is described with a respective hierarchical feature vector that is based on the region feature vectors of the respective parent and child regions. The hierarchical structures are classified according to a set of predefined classes with a hierarchical classifier component that is trained with hierarchical feature vectors of hierarchical structures of training images. The training images have semantic regions labeled according to the set of predefined classes. The input image is segmented into a plurality of semantic regions based on the classification of the hierarchical structures and optionally also on classification of the individual regions.

摘要翻译： 图像分割方法包括通过输入图像的无监督分割来生成区域的层级。 用表示该区域的相应区域特征向量描述每个区域。 识别分层结构，每个结构包括层次结构中的父区域及其各自的子区域。 使用基于相应父子区域和子区域的区域特征向量的相应分层特征向量来描述每个分层结构。 层次结构根据一组具有层次分类器组件的预定义类别进行分类，该层级分类器组件用训练图像的层次结构的分层特征向量进行训练。 训练图像具有根据预定义类的集合标记的语义区域。 基于层次结构的分类，并且还可以根据各个区域的分类，将输入图像分割成多个语义区域。

公开(公告)号：US08824797B2

公开(公告)日：2014-09-02

申请号：US13251459

申请日：2011-10-03

申请人： Neda Salamati ,  Diane Larlus ,  Gabriela Csurka ,  Craig Saunders

发明人： Neda Salamati ,  Diane Larlus ,  Gabriela Csurka ,  Craig Saunders

IPC分类号： G06K9/34 ,  G06K9/62 ,  G06T7/00

CPC分类号： G06T7/0087 ,  G06T7/143 ,  G06T7/162 ,  G06T2207/10048 ,  G06T2207/20072 ,  G06T2207/20076 ,  G06T2207/20081

摘要： A method for segmenting an image includes extracting unary potentials for pixels of the input image. These can be based for each of a set of possible labels, on information for a first channel in the image, such as in the visible range of the spectrum. Pairwise potentials are extracted for neighboring pairs of pixels of the image. These can be based on information for a second channel in the image, such as in the infrared range of the spectrum. An objective function is optimized over pixels of the input image to identify labels for the pixels. The objective function is based on a combination of ones of the extracted unary and pairwise potentials. The image is then segmented, based on the identified pixel labels. The method and system can provide an improvement in segmentation over methods which use only the visible information.

摘要翻译： 用于分割图像的方法包括提取输入图像的像素的一元电位。 这些可以基于用于图像中的第一通道的信息的一组可能标签中的每一个，例如在频谱的可见范围内。 为图像的相邻像素对提取成对电位。 这些可以基于图像中的第二通道的信息，例如在光谱的红外范围中。 在输入图像的像素上优化目标函数，以识别像素的标签。 目标函数基于提取的一元和成对电位的组合。 然后基于所识别的像素标签来分割图像。 该方法和系统可以提供对仅使用可见信息的方法进行分割的改进。

公开(公告)号：US08489585B2

公开(公告)日：2013-07-16

申请号：US13331096

申请日：2011-12-20

申请人： Diane Larlus ,  Florent Perronnin

发明人： Diane Larlus ,  Florent Perronnin

IPC分类号： G06F7/00 ,  G06F17/30

CPC分类号： G06F17/30017

摘要： A document processing system and method are disclosed. In the method local scores are incrementally computed for document samples, based on local features extracted from the respective sample. A global score is estimated for the document based on the local scores currently computed, i.e., on fewer than all document samples. A confidence in a decision for the estimated global score is computed. The computed confidence is based on the local scores currently computed and, optionally, the number of samples used in computing the estimated global score. A classification decision, such as a categorization or retrieval decision for the document is output, based on the estimated score when the computed confidence in the decision reaches a threshold value.

摘要翻译： 公开了一种文件处理系统和方法。 在该方法中，基于从相应样本提取的局部特征，对文档样本递增地计算局部分数。 基于当前计算的本地分数（即，少于所有文档样本）为文档估计全局分数。 计算对估计全局分数的决定的信心。 所计算的置信度基于当前计算的局部分数，以及可选地，用于计算估计全局得分的样本数。 当计算出的对判定的置信度达到阈值时，基于估计的分数来输出诸如文档的分类或检索决定的分类决定。

公开(公告)号：US20130159292A1

公开(公告)日：2013-06-20

申请号：US13331096

申请日：2011-12-20

申请人： Diane Larlus ,  Florent Perronnin

发明人： Diane Larlus ,  Florent Perronnin

IPC分类号： G06F17/30

CPC分类号： G06F17/30017

摘要： A document processing system and method are disclosed. In the method local scores are incrementally computed for document samples, based on local features extracted from the respective sample. A global score is estimated for the document based on the local scores currently computed, i.e., on fewer than all document samples. A confidence in a decision for the estimated global score is computed. The computed confidence is based on the local scores currently computed and, optionally, the number of samples used in computing the estimated global score. A classification decision, such as a categorization or retrieval decision for the document is output, based on the estimated score when the computed confidence in the decision reaches a threshold value.

摘要翻译： 公开了一种文件处理系统和方法。 在该方法中，基于从相应样本提取的局部特征，对文档样本递增地计算局部分数。 基于当前计算的本地分数（即，少于所有文档样本）为文档估计全局分数。 计算对估计全局分数的决定的信心。 所计算的置信度基于当前计算的局部分数，以及可选地，用于计算估计全局得分的样本数。 当计算出的对判定的置信度达到阈值时，基于估计的分数来输出诸如文档的分类或检索决定的分类决定。

公开(公告)号：US20140037198A1

公开(公告)日：2014-02-06

申请号：US13567309

申请日：2012-08-06

申请人： Diane Larlus-Larrondo ,  Rui Hu ,  Craig Saunders ,  Gabriela Csurka

发明人： Diane Larlus-Larrondo ,  Rui Hu ,  Craig Saunders ,  Gabriela Csurka

IPC分类号： G06K9/34 ,  G06K9/62

CPC分类号： G06K9/00 ,  G06T7/11 ,  G06T7/162 ,  G06T7/194 ,  G06T2207/20076

摘要： An image segmentation method includes generating a hierarchy of regions by unsupervised segmentation of an input image. Each region is described with a respective region feature vector representative of the region. Hierarchical structures are identified, each including a parent region and its respective child regions in the hierarchy. Each hierarchical structure is described with a respective hierarchical feature vector that is based on the region feature vectors of the respective parent and child regions. The hierarchical structures are classified according to a set of predefined classes with a hierarchical classifier component that is trained with hierarchical feature vectors of hierarchical structures of training images. The training images have semantic regions labeled according to the set of predefined classes. The input image is segmented into a plurality of semantic regions based on the classification of the hierarchical structures and optionally also on classification of the individual regions.

摘要翻译： 图像分割方法包括通过输入图像的无监督分割来生成区域的层级。 用表示该区域的相应区域特征向量来描述每个区域。 识别分层结构，每个结构包括层次结构中的父区域及其各自的子区域。 使用基于相应父子区域和子区域的区域特征向量的相应分层特征向量来描述每个分层结构。 层次结构根据一组具有层次分类器组件的预定义类别进行分类，该层级分类器组件用训练图像的层次结构的分层特征向量进行训练。 训练图像具有根据预定义类的集合标记的语义区域。 基于层次结构的分类，并且还可以根据各个区域的分类，将输入图像分割成多个语义区域。

公开(公告)号：US20130084007A1

公开(公告)日：2013-04-04

申请号：US13251459

申请日：2011-10-03

申请人： Neda Salamati ,  Diane Larlus ,  Gabriela Csurka ,  Craig Saunders

发明人： Neda Salamati ,  Diane Larlus ,  Gabriela Csurka ,  Craig Saunders

IPC分类号： G06K9/34

CPC分类号： G06T7/0087 ,  G06T7/143 ,  G06T7/162 ,  G06T2207/10048 ,  G06T2207/20072 ,  G06T2207/20076 ,  G06T2207/20081

摘要： A method for segmenting an image includes extracting unary potentials for pixels of the input image. These can be based for each of a set of possible labels, on information for a first channel in the image, such as in the visible range of the spectrum. Pairwise potentials are extracted for neighboring pairs of pixels of the image. These can be based on information for a second channel in the image, such as in the infrared range of the spectrum. An objective function is optimized over pixels of the input image to identify labels for the pixels. The objective function is based on a combination of ones of the extracted unary and pairwise potentials. The image is then segmented, based on the identified pixel labels. The method and system can provide an improvement in segmentation over methods which use only the visible information.

摘要翻译： 用于分割图像的方法包括提取输入图像的像素的一元电位。 这些可以基于用于图像中的第一通道的信息的一组可能标签中的每一个，例如在频谱的可见范围内。 为图像的相邻像素对提取成对电位。 这些可以基于图像中的第二通道的信息，例如在光谱的红外范围中。 在输入图像的像素上优化目标函数，以识别像素的标签。 目标函数基于提取的一元和成对电位的组合。 然后基于所识别的像素标签来分割图像。 该方法和系统可以提供对仅使用可见信息的方法进行分割的改进。