检索结果-内蒙古大学图书馆

IEEE International Symposium on Circuits and systems

作者： Li Chen Miska M. Hannuksela Houqiang Li CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China Hefei China Nokia Technologies Tampere Finland

ISBN: (纸本)9781479983926

The scalable and multiview extensions of the High Efficiency Video Coding share the same high-level syntax coding structure. For the scalable extension, the motion field of the inter-layer reference picture is modified through Motion Field Mapping before used for motion vector prediction. However, the motion field of the inter-layer reference picture is used without modification in the multiview extension. In this paper, a disparity-compensated inter-layer motion prediction is proposed for multiview video coding to achieve disparity compensation in inter-layer motion prediction using the scaled reference layer offset. The experimental results show that comparing with the multiview extension anchor, the proposed method achieves an average of 1.1% bitrate reduction.

关键词： Video coding motion vector Scalability

来源：评论

学校读者我要写书评

暂无评论

Offline and Online Optical Flow Enhancement for Deep Video Compression

arXiv

引用

arXiv 2023年

作者： Tang, Chuanbo Sheng, Xihua Li, Zhuoyuan Zhang, Haotian Li, Li Liu, Dong CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System Department of Electronic Engineering and Information Science University of Science and Technology of China Hefei230027 China

Video compression relies heavily on exploiting the temporal redundancy between video frames, which is usually achieved by estimating and using the motion information. The motion information is represented as optical flows in most of the existing deep video compression networks. Indeed, these networks often adopt pre-trained optical flow estimation networks for motion estimation. The optical flows, however, may be less suitable for video compression due to the following two factors. First, the optical flow estimation networks were trained to perform inter-frame prediction as accurately as possible, but the optical flows themselves may cost too many bits to encode. Second, the optical flow estimation networks were trained on synthetic data, and may not generalize well enough to real-world videos. We address the twofold limitations by enhancing the optical flows in two stages: offline and online. In the offline stage, we fine-tune a trained optical flow estimation network with the motion information provided by a traditional (non-deep) video compression scheme, e.g. H.266/VVC, as we believe the motion information of H.266/VVC achieves a better rate-distortion tradeoff. In the online stage, we further optimize the latent features of the optical flows with a gradient descent-based algorithm for the video to be compressed, so as to enhance the adaptivity of the optical flows. We conduct experiments on a state-of-the-art deep video compression scheme, DCVC. Experimental results demonstrate that the proposed offline and online enhancement together achieves on average 12.8% bitrate saving on the tested videos, without increasing the model or computational complexity of the decoder side. Copyright © 2023, The Authors. All rights reserved.

关键词： Image coding

来源：评论

学校读者我要写书评

暂无评论

Topology-based automatic 3D modeling from multiple images

Topology-based automatic 3D modeling from multiple images

引用

International Conference on Wireless Communications and Signal processing (WCSP)

作者： Long Zhang Xuejin Chen CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China Hefei P. R. China

3D reconstruction from multiple-view images has drawn a lot of attentions in computer graphics and computer vision communities. Traditional techniques usually end at discrete 3D point clouds computed from feature correspondence. However, geometric structure remains ambiguous in these unstructured point clouds, especially for man-made objects like buildings, indoor scenes. This paper proposes a new method to automatically reconstruct the main geometric structure of the scene composed of planar faces. First, dense 3D point clouds are reconstructed by applying patch-based multi-view stereo (PMVS) algorithm [1]. Then 3D planar primitives are extracted using a RANSAC-based approach [2]. We present a novel method to analyze the adjacency relations of the planar primitives to estimate the 3D intersection lines on the corresponding faces. Junctions and polygonal faces are computed from the 3D intersection lines along with complementary image information to compose the topology structure. Finally, texture for each face is extracted from the image under the best view. Experimental results demonstrate the feasibility of our system by successfully reconstructing the main structure of a wide range of scenes and constructing a texture-mapped piecewise-planar 3D model from images in multiple views.

关键词： Three-dimensional displays Image reconstruction Solid modeling Computational modeling Junctions Topology Surface reconstruction

来源：评论

学校读者我要写书评

暂无评论

Networked Predictive Control Based On The Round-Trip Delay

Networked Predictive Control Based On The Round-Trip Delay

引用

第三十四届中国控制会议

作者： Wang Binglin Kang Yu Yin Baoqun Xie Zhen Department of Automation University of Science and Technology of China The State Key Laboratory of Fire Science University of Science and Technology of China and Key Laboratory of Technology in Geo-Satial Information Processing and Application System Chinese Academy of Science School of Electrical Engineering and Automation Hefei University of Technology

ISBN: (纸本)9781467374439

Network predictive control is effectively in dealing with time-delay, data dropouts, packets disorders of Network Control System. This paper studies the stochastic stability of a affine nonlinear system within random time delay. The system controller is consist of a control prediction generator(CPG) at the controller side and a network delay compensator(NDC) at the plant side. Different from the previous literatures that model the random delay of the feedback channel and the forward channel as Markov chain respectively, the model in this paper is based on the round-trip delay, which making the predictive control process more concise but without losing generality. Then, this paper gives the necessary and sufficient conditions for stochastic stability of the closed-loop system. Finally, a numerical example is given to demonstrate the effectiveness of the proposed method.

关键词： Predictive Control,Affine System,Markov Process,Stochastic Stability

来源：评论

学校读者我要写书评

暂无评论

Sasl: Saliency-adaptive sparsity learning for neural network acceleration

arXiv

引用

arXiv 2020年

作者： Shi, Jun Xu, Jianfeng Tasaka, Kazuyuki Chen, Zhibo CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China Hefei230027 China KDDI Research Japan

Accelerating the inference speed of CNNs is critical to their deployment in real-world applications. Among all the pruning approaches, those implementing a sparsity learning framework have shown to be effective as they learn and prune the models in an end-to-end data-driven manner. However, these works impose the same sparsity regularization on all filters indiscriminately, which can hardly result in an optimal structuresparse network. In this paper, we propose a Saliency-Adaptive Sparsity Learning (SASL) approach for further optimization. A novel and effective estimation of each filter, i.e., saliency, is designed, which is measured from two aspects: The importance for the prediction performance and the consumed computational resources. During sparsity learning, the regularization strength is adjusted according to the saliency, so our optimized format can better preserve the prediction performance while zeroing out more computation-heavy filters. The calculation for saliency introduces minimum overhead to the training process, which means our SASL is very efficient. During the pruning phase, in order to optimize the proposed data-dependent criterion, a hard sample mining strategy is utilized, which shows higher effectiveness and efficiency. Extensive experiments demonstrate the superior performance of our method. Notably, on ILSVRC-2012 dataset, our approach can reduce 49.7% FLOPs of ResNet-50 with very negligible 0.39% top-1 and 0.05% top-5 accuracy degradation. Copyright © 2020, The Authors. All rights reserved.

关键词： Convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

Study on the three dimension imaging methods of fully-polarised array InSAR

Study on the three dimension imaging methods of fully-polari...

引用

IET International Radar Conference (IRC 2023)

作者： Shujie Song Xiaolan Qiu Songtao Shangguan Key Laboratory of Technology in Geo-spatial Information Processing and Application System Chinese Academy of Sciences Beijing People's Republic of China Suzhou Key Laboratory of Microwave Imaging Processing and Application Technology Suzhou People's Republic of China

ISBN: (数字)9781837240982

Fully-polarised (FP) array interferometric Synthetic Aperture Radar (FP-Array-InSAR) is an important technology in three-dimensional (3D) reconstruction and image interpretation of various scattering mechanisms (SMs) by exploiting the structural and polarisation properties of the targets. There has been a solid foundation for multi-baseline PolInSAR, but lack of systematic comparison and analysis for FP-Array-InSAR. Relying on the UAV-borne FP-Array-InSAR system developed by our research team, this paper applies two 3D imaging methods, one based on polarisation decomposition and the other based on multi-baseline (MB) polarimetric coherence optimization, to four-channel FP-array-InSAR datasets from Suzhou, China. Finally, this paper compares and analyses the experimental results of these two methods, and then gives their applicable scenarios. Such comparative experiments are positive for the rational selection and application of different methods in FP-array-InSAR.

关键词：

来源：评论

学校读者我要写书评

暂无评论

A Method for Object Occlusion Handling Based on Multilevel Track Queues

A Method for Object Occlusion Handling Based on Multilevel T...

引用

International Conference on Multimedia and Signal processing (CMSP)

作者： Biao Jin Wenlong Hu Yu Zhang Institute of Electronic Key Laboratory of Technology in Geo-spatial Information Processing and Application System CAS Graduate School Chinese Academy and Sciences Beijing China Institute of Electronic Key Laboratory of Technology in Geo-spatial Information Processing and Application System Chinese Academy and Sciences Beijing China China Electrical Power Research Institute Beijing China

Multi-object tracking is important in many computer vision applications. The major difficulties are due to inter-object or scene occlusion and data association. In this paper, we present a method to automatically detect and track multiple objects in complicated scenario from a single camera that is stationary. The proposed detecting method can prevent the moving objects updated to the background and can restrain ghosts. The moving objects are modeled fusing the hue and edge features. The system achieves tracking the object using the proposed associated strategy based on multilevel track queues. Experiments show the effectiveness of our approach under different conditions with complete occlusions.

关键词： Tracking Image edge detection Pixel Cameras Videos Computational modeling Algorithm design and analysis

来源：评论

学校读者我要写书评

暂无评论

Image Deblocking via Group Sparsity Optimization

Image Deblocking via Group Sparsity Optimization

引用

IEEE International Symposium on Circuits and systems

作者： Zhenbo Lu Houqiang Li Weiping Li CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System Department of Electronic Engineering and Information Science University of Science and Technology of China Heifei Anhui China

ISBN: (纸本)9781479983926

Block-wise compressed image often suffers from the blocking artifacts. In this paper, we propose a novel deblocking scheme for compressed image, by combining image's sparse property and its self-similarity together, called group sparsity optimization. Instead of processing each image patch individually, in the proposed scheme, similar patches in one group are required to be well-represented on learned dictionary collaboratively, using group sparsity regularization. The group sparsity not only imposes every patch's representation to be sparse, bus also requires patches' coefficients in the group share the similar pattern. The experiment results on standard test images demonstrate that our scheme can improve the PSNR of the compressed images by an average of 1.25 dB, and outperform state of the art deblocking approaches.

关键词： Image Deblocking Image Restoration JPEG compression

来源：评论

学校读者我要写书评

暂无评论

Extracting Named Entity Using Entity Labeling in geological Text Using Deep Learning Approach

引用

Journal of Earth Science 2023年第5期34卷 1406-1417页

作者： Qinjun Qiu Miao Tian Zhong Xie Yongjian Tan Kai Ma Qingfang Wang Shengyong Pan Liufeng Tao State Key Laboratory of Geo-Information Engineering and Key Laboratory of Surveying and Mapping Science and Geospatial Information Technology of MNR Chinese Academy of Surveying and MappingBeijing 100036China Key Laboratory of Spatial-Temporal Big Data Analysis and Application of Natural Resources in Megacities MNRShanghai 20063China Key Laboratory of Geological Survey and Evaluation of Ministry of Education China University of GeosciencesWuhan 430074China Hubei Key Laboratory of Intelligent Vision Based Monitoring for Hydroelectric Engineering China Three Gorges UniversityYichang 443002China College of Computer and Information Technology China Three Gorges UniversityYichang 443002China Chengdu Geological Environment Monitoring Station Chengdu 610036China Wuhan Zondy Cyber Science&Technology Co. Ltd.Wuhan 430074China

Artificial intelligence(AI) is the key to mining and enhancing the value of big data, and knowledge graph is one of the important cornerstones of artificial intelligence, which is the core foundation for the integration of statistical and physical representations. Named entity recognition is a fundamental research task for building knowledge graphs, which needs to be supported by a high-quality corpus, and currently there is a lack of high-quality named entity recognition corpus in the field of geology, especially in Chinese. In this paper, based on the conceptual structure of geological ontology and the analysis of the characteristics of geological texts, a classification system of geological named entity types is designed with the guidance and participation of geological experts, a corresponding annotation specification is formulated, an annotation tool is developed, and the first named entity recognition corpus for the geological domain is annotated based on real geological reports. The total number of words annotated was 698 512 and the number of entities was 23 345. The paper also explores the feasibility of a model pre-annotation strategy and presents a statistical analysis of the distribution of technical and term categories across genres and the consistency of corpus annotation. Based on this corpus, a Lite Bidirectional Encoder Representations from Transformers(ALBERT)-Bi-directional Long Short-Term Memory(BiLSTM)-Conditional Random Fields(CRF) and ALBERT-BiLSTM models are selected for experiments, and the results show that the F1-scores of the recognition performance of the two models reach 0.75 and 0.65 respectively, providing a corpus basis and technical support for information extraction in the field of geology.

关键词： ontology geological reports named entity recognition geological corpus construction semi-automated annotation platforms deep learning

来源：评论

学校读者我要写书评

暂无评论

Image-to-image translation with multi-path consistency regularization

arXiv

引用

arXiv 2019年

作者： Lin, Jianxin Xia, Yingce Wang, Yijun Qin, Tao Chen, Zhibo CAS Key Laboratory of Technology in Geo-spatial Information Processing and Application System University of Science and Technology of China University of Science and Technology of China Microsoft Research Asia

Image translation across different domains has attracted much attention in both machine learning and computer vision communities. Taking the translation from source domain Ds to target domain Dt as an example, existing algorithms mainly rely on two kinds of loss for training: One is the discrimination loss, which is used to differentiate images generated by the models and natural images;the other is the reconstruction loss, which measures the difference between an original image and the reconstructed version through Ds → Dt → Ds translation. In this work, we introduce a new kind of loss, multi-path consistency loss, which evaluates the differences between direct translation Ds → Dt and indirect translation Ds → Da → Dt with Da as an auxiliary domain, to regularize training. For multi-domain translation (at least, three) which focuses on building translation models between any two domains, at each training iteration, we randomly select three domains, set them respectively as the source, auxiliary and target domains, build the multi-path consistency loss and optimize the network. For two-domain translation, we need to introduce an additional auxiliary domain and construct the multi-path consistency loss. We conduct various experiments to demonstrate the effectiveness of our proposed methods, including face-toface translation, paint-to-photo translation, and deraining/de-noising translation. Copyright © 2019, The Authors. All rights reserved.

关键词： Iterative methods

来源：评论

学校读者我要写书评

暂无评论

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案：

请选择收藏分类：

通借通还

建议与咨询 留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

时间限定

文献类型

馆藏选择

核心期刊

语言

文献类型

帮助

文字说明：

检索规则说明：

检索范例：

分类表

所选分类

限定检索结果

文献类型

馆藏范围

日期分布

学科分类号

主题

机构

作者

语言

请选择保存的检索档案： 新增检索档案 确定 取消

请选择收藏分类： 新增自定义分类 确定 取消

通借通还

建议与咨询留下您的常用邮箱和电话号码，以便我们向您反馈解决方案和替代方法

请选择保存的检索档案：

请选择收藏分类：