Beihang University

Xiaowu Chen

Qinping Zhao

Dongqing Zou

Bin Zhou

Jia Qi Li

Qiang Fu

Xiaogang Wang

Yu Zhang

Jianwei Li

Feng Ding

Xiaotian Wang

Changqun Xia

Kan Guo

Bo Gao

Feng Lu

Sijia Wen

Yafei Song

Anlin Zheng

Xiaoyu Su

Chen Wang

Qing Ql Li

Lin Wang

Feixiang Lu

Yifan Zhao

Zongji Wang

Haiyue Fang

Wei Teng

Haokun Song

Guangying Cao

Lang Bi

Disclosed are a method and apparatus for adaptively constructing a three-dimensional indoor scenario, the method including: establishing an object association map corresponding to different scenario categories according to an annotated indoor layout; selecting a corresponding target indoor object according to room information inputted by a user and the object association map; generating a target indoor layout according to preset room parameters inputted by the user and the annotated indoor layout; and constructing a three-dimensional indoor scenario according to the target indoor object and the target indoor layout. The disclosed method and apparatus help improving the efficiency in constructing the three-dimensional scenario.

Method and apparatus for adaptively constructing three-dimensional indoor scenario

Embodiments of the present invention provide a method and a device for three-dimensional feature-embedded image object component-level semantic segmentation, the method includes: acquiring three-dimensional feature information of a target two-dimensional image; performing a component-level semantic segmentation on the target two-dimensional image according to the three-dimensional feature information of the target two-dimensional image and two-dimensional feature information of the target two-dimensional image. In the technical solution of the present application, not only the two-dimensional feature information of the image but also the three-dimensional feature information of the image are taken into consideration when performing the component-level semantic segmentation on the image, thereby improving the accuracy of the image component-level semantic segmentation.

Method and device for three-dimensional feature-embedded image object component-level semantic segmentation

Disclosed is a co-segmentation method and apparatus for a three-dimensional model set, which includes: obtaining a super patch set for the three-dimensional model set which includes at least two three-dimensional models, each of the three-dimensional models including at least two super patches; obtaining a consistent affinity propagation model according to a first predefined condition and a conventional affinity propagation model, the consistent affinity propagation model being constraint by the first predefined condition which is position information for at least two super patches that are in the super patch set and belong to a common three-dimensional model set; converting the consistent affinity propagation model into a consistent convergence affinity propagation model; clustering the super patch set through the consistent convergence affinity propagation model to generate a co-segmentation outcome for the three-dimensional model set. The disclosed three-dimensional model set co-segmentation method and apparatus improves consistency between three-dimensional model set co-segmentation outcomes.

Co-segmentation method and apparatus for three-dimensional model set

The disclosed provides a posture-guided cross-category method and device for combination modeling of cross-category 3D models, the method including: receiving a first posture model inputted by a user; calculating similarities between q first regions of the first posture model and q second regions of each of second posture models in a preset model database, respectively; where the model database includes a plurality of models partitioned into model components, and second posture models corresponding to the models; where the q first regions correspond to the q second regions one on one, where q is an integer greater than or equal to 2; selecting a corresponding plurality of model components of the q second regions of the second posture model according to the similarities; and combining the selected plurality of model components to generate a 3D model. The embodiment of the present application can combine cross-category models with large structural differences.

Posture-guided method and device for combination modeling of cross-category three-dimensional models

Present invention provides a ranking convolutional neural network constructing method and an image processing method and apparatus thereof. The ranking convolutional neural network includes a ranking layer that is configured to sort an output of a previous layer of the ranking layer, generate an output of the ranking layer according to the sorted output, and output the output of the ranking layer to a next layer of the ranking layer. Using the ranking convolutional neural network enables obtaining an output feature corresponding to the input feature image through automatic learning. Compared with prior art methods that obtain features through manual calculation, the method of the present invention is superior in terms of reflecting the objective laws contained by the patterns of the actual scene. When applied to the field of image processing, the method can significantly improve the effect of image processing.

Image dehazing method

The present application provides a method and an apparatus for parsing and processing a three-dimensional CAD model, where the method includes: determining three kinds of adjacency relation information for each component in the three-dimensional model; performing aggregation processing on all components of the three-dimensional model, and generating three part hypothesis sets for the three-dimensional model; performing voxelization expression processing on each part hypothesis in each part hypothesis set, and generating voxelization information for each part hypothesis; inputting voxelization information of all part hypotheses in each part hypothesis set into an identification model to obtain a confidence score and a semantic category probability distribution for each part hypothesis; and constructing, according to the confidence score and the semantic category probability distribution of each part hypothesis in the part hypothesis sets, a high-order conditional random field model, and obtaining a semantic category analysis result for each component in the three-dimensional model.

Method and apparatus for parsing and processing three-dimensional CAD model

Provided is a method for automatic constructing a three-dimensional indoor scenario with behavior-constrained, including: determining a room, parameter of the room and a first behavioral semantics according to entered information; obtaining a corresponding first object category according to the first behavioral semantics; determining a corresponding first reference layout and a three-dimensional model of objects in the room according to the first object category and the parameter of the room; and directing and generating a three-dimensional indoor scenario in the room according to the first reference layout and the three-dimensional model.

Method for automatic constructing three-dimensional indoor scenario with behavior-constrained

A method and an apparatus for extracting a saliency map are provided in the embodiment of the present application, the method includes: conducting first convolution processing, first pooling processing and normalization processing on an original image via a prediction model to obtain eye fixation information from the original image, where the eye fixation information is used for indicating a region at which human eye gaze; conducting second convolution processing and second pooling processing on the original image via the prediction model to obtain semantic description information from the original image; fusing the eye fixation information and the semantic description information via element-wise summation function; and conducting detection processing on the fused eye fixation information and semantic description information via the prediction model to obtain a saliency map from the original image. It is used for improving the efficiency of extracting the saliency map from image.

Methods and apparatus for image salient object detection

Provided is a method for salient object segmentation of an image by aggregating a multi-linear exemplar regressors, including: analyzing and summarizing visual attributes and features of a salient object and a non-salient object using background prior and constructing a quadratic optimization problem, calculating an initial saliency probability map, selecting a most trusted foreground and a background seed point, performing manifold preserving foreground propagation, generating a final foreground probability map; generating a candidate object set for the image via an objectness adopting proposal, using a shape feature, a foregroundness and an attention feature to characterize each candidate object, training the linear exemplar regressors for each training image to characterize a particular saliency pattern of the image; aggregating a plurality of linear exemplar regressors, calculating saliency values for the candidate object set of a test image, and forming an image salient object segmentation model capable of processing various complex scenarios.

Method for salient object segmentation of image by aggregating multi-linear exemplar regressors

The disclosure involves an image guided video semantic object segmentation method and apparatus, locate a target object in a sample image to obtain an object sample; extract a candidate region from each frame; match multiple candidate regions extracted from the each frame with the object sample to obtain a similarity rating of each candidate region; rank the similarity rating of each candidate region to select a predefined candidate region number of high rating candidate region ranked by the similarity; preliminarily segment a foreground and a background from the selected high rating candidate region; construct an optimization function for the preliminarily segmented foreground and background; solve the optimization function to obtain a optimal candidate region set; and propagate a preliminary foreground segmentation corresponding to the optimal candidate region to an entire video to obtain a semantic object segmentation of the input video.

Image guided video semantic object segmentation method and apparatus

Provided is a method and a system for detecting and segmenting primary video objects with neighborhood reversibility, including: dividing each video frame of a video into super pixel blocks; representing each super pixel block with visual features; constructing and training a deep neural network to predict the initial foreground value for each super pixel block in the spatial domain; constructing a neighborhood reversible matrix and transmitting the initial foreground value, constructing an iterative optimization problem and resolving the final foreground value in the temporal spatial domain; performing pixel level transformation on the final foreground value; optimizing the final foreground value for the pixel using morphological smoothing operations; determining whether the pixel belongs to the primary video objects according to the final foreground value. The present disclosure does not require to set a priori assumption for processing a video, and is especially suitable for big data sets including complicated scenarios.

Primary video objects with neighborhood reversibility

The present invention provides a method for video matting via sparse and low-rank representation, which firstly selects frames which represent video characteristics in input video as keyframes, then trains a dictionary according to known pixels in the keyframes, next obtains a reconstruction coefficient satisfying the restriction of low-rank, sparse and non-negative according to the dictionary, and sets the non-local relationship matrix between each pixel in the input video according to the reconstruction coefficient, meanwhile sets the Laplace matrix between multiple frames, obtains a video alpha matte of the input video, according to &#x3b1; values of the known pixels of the input video and &#x3b1; values of sample points in the dictionary, the non-local relationship matrix and the Laplace matrix; and finally extracts a foreground object in the input video according to the video alpha matte, therefore improving quality of the extracted foreground object.

Method for video matting via sparse and low-rank representation

Presently disclosed is a method for co-segmenting three-dimensional models represented by sparse and low-rank feature, comprising: pre-segmenting each three-dimensional model of a three-dimensional model class to obtain three-dimensional model patches for the each three-dimensional model; constructing a histogram for the three-dimensional model patches of the each three-dimensional model to obtain a patch feature vector for the each three-dimensional model; performing a sparse and low-rank representation to the patch feature vector for the each three-dimensional model to obtain a representation coefficient and a representation error of the each three-dimensional model; determining a confident representation coefficient for the each three-dimensional model according to the representation coefficient and the representation error of the each three-dimensional model; and clustering the confident representation coefficient of the each three-dimensional model to co-segment the each three-dimensional model respectively.

Method for co-segmentating three-dimensional models represented by sparse and low-rank feature

The invention provides a calculation method of line-of-sight direction based on analysis and match of iris contour in human eye image, including: a data driven method, for stable calculation of 3D line-of-sight direction via inputting human eye image to be matched with synthetic data of virtual eyeball appearance; two novel optimization matching criterions of eyeball appearance, which effectively reduce effects of uncontrollable factors, such as image scaling and noise on results; a joint optimization method, for the case of continuously shooting multiple human eye images, to further improve calculation accuracy. One application of the invention is virtual reality and human computer interaction which is under the principle that shooting eye images of a user and calculating line-of-sight direction of user to enable interaction with intelligent system interface or virtual realistic object. The invention can be widely used in training, games and entertainment, video surveillance, medical care and other fields.

Calculation method of line-of-sight direction based on analysis and match of iris contour in human eye image

The invention provides a data driven method for transferring indoor scene layout and color style, including: preprocessing images in an indoor image data set, which includes manually labeling semantic information and layout information; obtaining indoor layout and color rules on the data set by learning algorithms; performing object-level semantic segmentation on input indoor reference image, or performing object-level and component-level segmentations using color segmentation methods, to extract layout constraints and color constraints of reference images, associating the reference images with indoor 3D scene via the semantic information; constructing a graph model for indoor reference image scene and indoor 3D scene to express indoor scene layout and color; performing similarity measurement on the indoor scene and searching for similar images in the data set to obtain an image sequence with gradient layouts from reference images to input 3D scene; performing image-sequence-guided layout and color transfer generation for indoor 3D scene.

Growing community of inventors

Beijing, China

Xiaowu Chen