检索结果-内蒙古大学图书馆

IEEE visual communications and image processing (VCIP)

作者： Qiuming Liu Yichen Wang Ying Wei Lei Xie Changjian Zhu Ruoxuan Zhou School of Software Engineering Jiangxi Univ. of Sci. and Tech. Jiangxi China School of Comm. and Infor. Engineering Chongqing Univ. of Posts and Telecom. Chongqing China School of Electron. and Inform. Comm. Huazhong Univ. of Sci. and Tech. Wuhan China Nanchang Key lab. of Virtual Digital Factory and Cultural Communications Jiangxi China

ISBN: (纸本)9781665475938

The aerial light field (ALF) can render higher quality images of large-scale 3D scenes. In this paper, we apply the ALF technology to study the image captured and novel view rendering of unmanned aerial vehicle (UAV), which exists some problems, such as large scene and depth of field. First, we design an ALF sampling model using spectral analysis of Fourier theory. Based on the ALF sampling model, the exact expression of ALF spectrum is derived. By the spectral support of ALF, we analyze the influence of pitch angle on light field sampling and its bandwidth. Particularly, the bandwidth of ALF can be applied to determine the minimum sampling rate for UAV. Additionally, we design a reconstruction filter that is related to pitch angle to render novel views of UAV. Finally, our experiments show that our sampling and rendering methods can improve the rendering quality of UAV novel view rendering.

关键词： Analytical models Three-dimensional displays visual communication Bandwidth Rendering (computer graphics) Autonomous aerial vehicles Light fields

来源：评论

学校读者我要写书评

暂无评论

Self-supervised Continual Learning for Object Recognition in image Sequences 28th

Self-supervised Continual Learning for Object Recognition in...

引用

28th International conference on Neural Information processing

作者： Dai, Ruiqi Lefort, Mathieu Armetta, Frederic Guillermin, Mathieu Difiner, Stefan Univ Lyon INSA Lyon Lyon France Univ Claude Bernard Lyon 1 Univ Lyon Lyon France CNRS UMR5205 LIRIS Villeurbanne France Univ Catholique Lyon Lyon France

ISBN: (纸本)9783030923068;9783030923075

The autonomous learning of different objects in images, with a continual and unsupervised context, relies on detecting unknown objects and recognizing known ones based on the learned visual representation. Novelty detection is challenging because of the internal representation drifts of known objects not been seen for a long time. Most existing approaches either perform offline unsupervised learning on a large dataset, or continual supervised learning. Nevertheless, very few existing approaches propose unsupervised continual learning for object recognition. In this paper, we propose a new neural network-based approach for continually learning representations of objects from image sequences, that is able to autonomously detect novel objects and to recognize previously learned ones during training. It is based on a statistical test, performed on internal representations, adapted to counterbalance the concept drift, without storing any image. Experimental results show that our approach outperforms the state of the art on MNIST and Fashion-MNIST datasets. In particular, our approach avoids to over-segment the distribution of clusters, which artificially increases traditional indicators such as clustering accuracy.

关键词： Object recognition

来源：评论

学校读者我要写书评

暂无评论

Multi-stage Locally and Long-range Correlated Feature Fusion for Learned In-loop Filter in VVC

Multi-stage Locally and Long-range Correlated Feature Fusion...

引用

IEEE visual communications and image processing (VCIP)

作者： Birendra Kathariya Zhu Li Hongtao Wang Geert Van Der Auwera University of Missouri-Kansas City Kansas City USA Qualcomm San Diego USA

ISBN: (纸本)9781665475938

Versatile Video Coding (VVC)/H.266 is currently the state-of-the-art video coding standard with significant improvement in coding efficiency over its predecessor High Efficiency Video Coding (HEVC)/H.26S. Nonetheless, VVC is also block-based video coding technology where decoded pictures contain compression artifacts. In VVC, in-loop filters serve to suppress these compression artifacts. In this paper, convolution neural network (CNN) is utilized to better facilitate the suppression of compression artifacts over VVC. Nonetheless, our approach has uniqueness in obtaining better features by exploiting locally correlated spatial features in the pixel domain as well as long-range correlated spectral features in the discrete cosine transform (DCT) domain. In particular, we utilized CNN-features from DCT transformed input to extract high-frequency components and induce long-range correlation into the spatial CNN-features by employing multi-stage feature fusion. Our experimental result shows that the proposed approach achieves significant coding improvements up to 9.70% on average Bjantegaard Delta (BD)-Bitrate savings under AI configurations for luma (Y) components.

关键词： image coding Correlation visual communication Convolution Neural networks Transforms Feature extraction

来源：评论

学校读者我要写书评

暂无评论

Memory-Efficient Point Cloud Registration via Overlapping Region Sampling

arXiv

引用

arXiv 2024年

作者： Shimada, Tomoyasu Murasaki, Kazuhiko Sato, Shogo Nishimura, Toshihiko Yoshida, Taiga Tanida, Ryuichi NTT Corporation Japan

Recent advances in deep learning have improved 3D point cloud registration but increased graphics processing unit (GPU) memory usage, often requiring preliminary sampling that reduces accuracy. We propose an overlapping region sampling method to reduce memory usage while maintaining accuracy. Our approach estimates the overlapping region and intensively samples from it, using a k-nearest-neighbor (kNN) based point compression mechanism with multi layer perceptron (MLP) and transformer architectures. Evaluations on 3DMatch and 3DLoMatch datasets show our method outperforms other sampling methods in registration recall, especially at lower GPU memory levels. For 3DMatch, we achieve 94% recall with 33% reduced memory usage, with greater advantages in 3DLoMatch. Our method enables efficient large-scale point cloud registration in resource-constrained environments, maintaining high accuracy while significantly reducing memory requirements. This paper is accepted for IEEE International conference on visual communications and image processing 20241 © 2024, CC BY.

关键词： Graphics processing unit

来源：评论

学校读者我要写书评

暂无评论

ML-FDA: Meta-Learning via Feature Distribution Alignment for Few-Shot Learning

ML-FDA: Meta-Learning via Feature Distribution Alignment for...

引用

IEEE visual communications and image processing (VCIP)

作者： Yuewen Li Binghao Liu Shuchang Lyu Lijiang Chen Qi Zhao Wenquan Feng Department of Electronic and Information Engineering Beihang University Beijing China

ISBN: (纸本)9781665475938

Computer vision tasks suffer from the high cost of collecting large amounts of labeled data. Few-shot Learning (FSL) is a dominant approach to solve this problem because it provides an insight to learn the knowledge of novel categories with few training samples. In FSL task, Meta-learning and metric learning have achieved impressive results. However, the performance of this task is still limited by large intra-class variance and small inter-class distance caused by limited number of few samples. To solve this problem, In this paper, we propose a new method, which integrates meta-learning and metric learning techniques. Specifically, we first propose a feature representation module (FR) to construct representative support class prototypes and query features. Then, we design bias loss to minimize the bias between support and query samples. Furthermore, we design an intra-class loss to minimize the distance between query class prototype and each query sample. We denote this model as ML-FDA and validate it on standard few-shot classification benchmark datasets (MiniimageNet, CIFAR-FS, FC100). The results show that our method improves the performance over other same paradigm methods and achieves the best performance on most benchmarks. The ablation study and visulization analysis also demonstrate the effectiveness of our method.

关键词： Measurement Training Computer vision Costs visual communication image processing Prototypes

来源：评论

学校读者我要写书评

暂无评论

ACCR: Auto-labeling for Ancient Chinese Handwritten Characters Recognition on CNN

ACCR: Auto-labeling for Ancient Chinese Handwritten Characte...

引用

IEEE visual communications and image processing (VCIP)

作者： Peikun Wu Xin Yang Fuhao Guo Li Wang Cihan Ruan School of Mathematical Sciences Nankai University Tianjin China China Institute of Science and Technology Information Beijing China Department of Computer Science Engineering Santa Clara University Santa Clara U.S

ISBN: (纸本)9781665475938

Chinese Character Recognition(CCR) is a critical application of Optical Character Recognition(OCR), a vital area of pattern recognition. Research on CCR in the past decades mainly focused on the modern Chinese characters, but not on the ancient ones. Compared to modern Chinese characters, ancient characters are more diverse and multiple ancient characters can correspond to one modern character. When doing recognition, the unique features of ancient Chinese characters cause a significant amount of time on manual labeling. This paper proposes an automatic labeling algorithm based on a semi-supervised dictionary training neural network that drastically decreases human effort. We first created an offline training set as a dictionary including 8,226 Chinese characters from ancient documents in modern fonts. And put the set into the network. Then we recursively retrained the network on an unlabeled data set of about 1.3 million characters images segmented from ancient documents resulting in a very high accuracy rate of 98.96 %. This work is one part of our wide recognition of ancient documents with handwritten Chinese characters project.

关键词： Training Handwriting recognition image recognition Dictionaries visual communication Neural networks Manuals

来源：评论

学校读者我要写书评

暂无评论

Deep Learning Based EBCOT Source Symbol Prediction Technique for JPEG2000 image Compression Architecture

Deep Learning Based EBCOT Source Symbol Prediction Technique...

引用

IEEE International conference on visual communications and image processing (VCIP)

作者： Wang, I-Hsiang Ding, Jian-Jiun Natl Taiwan Univ Grad Inst Commun Engn Taipei Taiwan

ISBN: (纸本)9781728180687

In this work, an efficient and robust learning-based JPEG2000 architecture is proposed. It uses machine learning techniques for predicting and encoding the decision bit in the embedded block coding with optimized truncation (EBCOT) process. First, we apply non-locally weighted ridge regression to predict the quantized wavelet coefficients in the LL subband. Then, during the EBCOT process, we perform inter/intra subband prediction and inter/intra bit plane symbol prediction to estimate the activity of the decision bit using the deep learning architecture. Then, the binary prediction result is treated as an additional context and the decision bit is eventually coded using an advanced context-based adaptive binary arithmetic coder. Simulations show that the proposed framework provides the same visual quality as conventional codecs with as much as 30% bitrate savings.

关键词： image compression JPEG2000 machine learning deep learning adaptive arithmetic coding

来源：评论

学校读者我要写书评

暂无评论

Near-lossless Point Cloud Geometry Compression Based on Adaptive Residual Compensation

Near-lossless Point Cloud Geometry Compression Based on Adap...

引用

IEEE visual communications and image processing (VCIP)

作者： Dingquan Li Jing Wang Ge Li Peng Cheng Laboratory Shenzhen China Peking University Shenzhen Graduate School Shenzhen China

ISBN: (纸本)9781665475938

Point cloud compression (PCC) is a crucial enabler for immersive multimedia applications since point cloud is one of the most primitive forms for representing 3D scenes and objects. Recently, some approaches are proposed to improve the average reconstruction quality of octree-based Geometry-based Point Cloud Compression (G-PCC). However, it is noticed that these approaches suffer considerable loss in terms of point-to-point (D1) Hausdorff distance when compared to G-PCC (octree). Here we introduce a near-lossless point cloud geometry compression method based on adaptive residual compensation by adding and removing points with large errors. It allows controlling of D1 Hausdorff (D1h) distance and maintains a great improvement in average reconstruction performance over G-PCC. Experimental results verify the effectiveness of our method, where our method achieves an average of 78.5% D1 and 11.4% D1h Bjontegaard-delta bitrate savings over the octree-based G-PCC on solid point clouds of the MPEG Cat1A dataset.

关键词： Point cloud compression Geometry image coding Three-dimensional displays visual communication Octrees Bit rate

来源：评论

学校读者我要写书评

暂无评论

History-parameter-based Affine Model Inheritance

History-parameter-based Affine Model Inheritance

引用

IEEE visual communications and image processing (VCIP)

作者： Kai Zhang Li Zhang Zhipin Deng Na Zhang Yang Wang Bytedance Inc.

ISBN: (纸本)9781665475938

In VVC, affine motion compensation (AMC) is a powerful coding tool to address non-translational motion, while history-based motion vector prediction (HMVP) is an efficient approach to compress motion vectors. However, HMVP was designed for translational motion vectors, without considering control point motion vectors (CPMV) for AMC. This paper presents a method of history-parameter-based affine model inheritance (HAMI), to utilize history information to represent CPMV more efficiently. With HAMI, affine parameters of previously affine-coded block are stored in a first history-parameter table (HPT). New affine-merge, affine motion vector prediction candidates and regular-merge candidates can be constructed with affine parameters fetched from the first HPT and base MVs fetched from neighbouring blocks in a “base-parameter-decoupled” way. New affine merge candidates can also be generated in a “base-parameter-coupled” way from a second HPT, which stores base MV information together with corresponding affine parameters. Besides, pair-wised affine merge candidates are generated by two existing affine merge candidates. Experimental results show that HAMI provides an average BD-rate saving about 0.34 % with a negligible change on the running time, compared with ECM-3.1 in random access configurations. HAMI has been adopted into ECM.

关键词： image coding visual communication Simulation Motion compensation Encoding Complexity theory History

来源：评论

学校读者我要写书评

暂无评论

Azimuth Adjustment Considering LiDAR Calibration for the Predictive Geometry Compression in G-PCC

Azimuth Adjustment Considering LiDAR Calibration for the Pre...

引用

IEEE visual communications and image processing (VCIP)

作者： Youguang Yu Wei Zhang Fuzheng Yang School of Telecommunications Engineering Xidian University China Artificial Intelligence Research Center Peng Cheng Laboratory China

ISBN: (纸本)9781665475938

Point clouds captured by spinning Light Detection And Ranging (LiDAR) devices have played a significant role in many applications. To efficiently store and transmit such a huge amount of data, MPEG designed the Geometry-based Point Cloud Compression (G-PCC) standard, which includes a dedicated Pre-dictive Profile for LiDAR point clouds. In this scheme, Cartesian coordinates are mapped to the spherical representation using the elevation-related LiDAR calibration parameters to better characterize the spherical acquisition pattern of the spinning LiDAR device. As such, stronger spatial correlations exist among neighbouring nodes in the predictive structure, resulting in higher compression efficiency. However, it should be mentioned that the azimuth-related calibration parameters, which are unused, also impact the accuracy and correctness of the mapped spherical coordinates. In this paper, an azimuth adjustment method is proposed taking into account this impact. Experimental results show that the proposed azimuth adjustment can consistently improve the coding efficiency of G-PCC. Furthermore, a LiDAR calibration parameter estimation method is proposed in case the azimuth-related parameters are absent. Results show that the proposed calibration parameter estimation can precisely approximate the ground truth.

关键词： Point cloud compression Geometry Laser radar image coding Parameter estimation Azimuth visual communication

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：