检索结果-内蒙古大学图书馆

Plug-and-Play Split Gibbs Sampler: Embedding Deep Generative Priors in Bayesian Inference

ieee TRANSACTIONS ON image processing 2024年 33卷 3496-3507页

作者： Coeurdoux, Florentin Dobigeon, Nicolas Chainais, Pierre Univ Toulouse CNRS IRIT INP ENSEEIHT F-31071 Toulouse France Univ Lille Ctr Rech Informat Signal & Automat Lille CRIStAL CNRS Cent LilleUMR 9189 F-59000 Lille France

This paper introduces a stochastic plug-and-play (PnP) sampling algorithm that leverages variable splitting to efficiently sample from a posterior distribution. The algorithm based on split Gibbs sampling (SGS) draws inspiration from the half quadratic splitting method (HQS) and the alternating direction method of multipliers (ADMM). It divides the challenging task of posterior sampling into two simpler sampling problems. The first problem depends on the likelihood function, while the second is interpreted as a Bayesian denoising problem that can be readily carried out by a deep generative model. Specifically, for an illustrative purpose, the proposed method is implemented in this paper using state-of-the-art diffusion-based generative models. Akin to its deterministic PnP-based counterparts, the proposed method exhibits the great advantage of not requiring an explicit choice of the prior distribution, which is rather encoded into a pre-trained generative model. However, unlike optimization methods (e.g., PnP-ADMM and PnP-HQS) which generally provide only point estimates, the proposed approach allows conventional Bayesian estimators to be accompanied by confidence intervals at a reasonable additional computational cost. Experiments on commonly studied image processing problems illustrate the efficiency of the proposed sampling strategy. Its performance is compared to recent state-of-the-art optimization and sampling methods.

关键词： Noise reduction Stochastic processes Inverse problems Data models Bayes methods Task analysis Kernel Bayesian inference plug-and-play prior deep generative model diffusion-based model Markov chain Monte Carlo inverse problem

来源：评论

学校读者我要写书评

暂无评论

Formal photograph compression in JPEG2000 coding

Formal photograph compression in JPEG2000 coding

引用

International symposium on intelligence Information processing and Trusted Computing

作者： Hou, Jun Chen, Wen Yang, Jing School of Optical Electrical and Computer Engineering University of Shanghai for Science and Technology Shanghai 200093 China Institute of Communication and Image Processing Shanghai Jiaotong University Shanghai China Maritime University Shanghai 200135 China

ISBN: (纸本)9780769541969

A formal facial image compression algorithm is proposed in this paper. Object segmentation techniques is used to locate the facial region firstly. An adapted version of the original Viola-Jones face detector is used to locate eyes and mouth. The position of chin is estimated and then checked according to the curve in the edge image. Secondly, the background is reassigned to decrease unvaluable information in final bitstream. Finally, the facial region is compressed using the region of interest coding technique in JPEG2000. The produced bitstream is totally compatible with the JPEG2000 standard one. As a result, the image compressed by the proposal algorithm could be delivered widely and browsed by many picture tools that support JPEG2000 format. Experiments demonstrate the efficiency of the method. © 2010 ieee.

关键词： image compression

来源：评论

学校读者我要写书评

暂无评论

Compressive Hyperspectral Imaging With Side Information

引用

ieee JOURNAL OF SELECTED TOPICS IN signal processing 2015年第6期9卷 964-976页

作者： Yuan, Xin Tsai, Tsung-Han Zhu, Ruoyu Llull, Patrick Brady, David Carin, Lawrence Duke Univ Dept Elect & Comp Engn Durham NC 27708 USA

A blind compressive sensing algorithm is proposed to reconstruct hyperspectral images from spectrally-compressed measurements. The wavelength-dependent data are coded and then superposed, mapping the three-dimensional hyperspectral datacube to a two-dimensional image. The inversion algorithm learns a dictionary in situ from the measurements via global-local shrinkage priors. By using RGB images as side information of the compressive sensing system, the proposed approach is extended to learn a coupled dictionary from the joint dataset of the compressed measurements and the corresponding RGB images, to improve reconstruction quality. A prototype camera is built using a liquid-crystal-on-silicon modulator. Experimental reconstructions of hyperspectral datacubes from both simulated and real compressed measurements demonstrate the efficacy of the proposed inversion algorithm, the feasibility of the camera and the benefit of side information.

关键词： Compressive sensing hyperspectral image side information Bayesian shrinkage dictionary learning blind compressive sensing computational photography coded aperture snapshot spectral imaging (CASSI) spatial light modulation

来源：评论

学校读者我要写书评

暂无评论

LOCALIZE HEAVILY OCCLUDED HUMAN FACES VIA DEEP SEGMENTATION 23

LOCALIZE HEAVILY OCCLUDED HUMAN FACES VIA DEEP SEGMENTATION

引用

23rd ieee International Conference on image processing (ICIP)

作者： Zhang, Kaihao Huang, Yongzhen He, Ran Wu, Hong Wang, Liang Univ Elect Sci & Technol China Sch Engn & Comp Sci Stat Machine Intelligence & LEarning Chengdu 611731 Sichuan Peoples R China Chinese Acad Sci Ctr Excellence Brain Sci & Intelligence Technol Inst Automat Natl Lab Pattern Recognit Beijing 100190 Peoples R China

ISBN: (纸本)9781467399616

Localizing heavily occluded human faces is a challenging problem in facial detection. Previous methods mainly employ sliding windows by determining whether windows include human faces. In this paper, we provide a novel segmentation based perspective for heavily occluded face localization with deep convolutional neural networks (CNN). Our model takes an image as input without complicated pre-processing. After several convolutional layers, fully-connected layers and a softmax classifier, we can predict the labels of all pixels in an image, which is the key to localize heavily occluded human faces. Finally, we search a minimal rectangle to localize the human face. Our detector needs neither complex pre-processing nor the time-consuming sliding window. Besides, we use a single model to localize faces to further alleviate computational complexity. Experimental results show that our proposed method is a very effective way to localize heavily occluded human face.

关键词： facial localization heavily occluded faces deep convolutional neural networks

来源：评论

学校读者我要写书评

暂无评论

SIMULTANEOUS INSTANCE ANNOTATION AND CLUSTERING IN MULTI-INSTANCE MULTI-LABEL LEARNING 25

SIMULTANEOUS INSTANCE ANNOTATION AND CLUSTERING IN MULTI-INS...

引用

ieee International Workshop on Machine Learning for signal processing

作者： Pham, Anh T. Raich, Raviv Fern, Xiaoli Z. Oregon State Univ Sch EECS Corvallis OR 97331 USA

ISBN: (纸本)9781467374545

Multi-instance multi-label learning (MIML) is a framework that addresses label ambiguity when data contains bags, each bag contains instances, and a bag label set is provided for each bag. Instance annotation in the MIML setting is the problem of finding an instance level classifier given training data consisting of labeled bags of instances. Current approaches for instance annotation mainly focus on identifying a class label for each instance without considering inner clusters within each class. Simultaneously learning to annotate and cluster may not only yield better model fit but also help to discovery cluster structure inside each class for future investigation. This paper addresses the challenge of simultaneously annotating and clustering by proposing a graphical model that takes into account inner clusters within each class. An expectation maximization inference based on maximum likelihood is proposed for the model. Results on bird song, image annotation, and two synthetic datasets illustrate the effectiveness of the proposed framework compared to current state-of-the-art approaches.

关键词： expectation-maximisation algorithm inference mechanisms learning (artificial intelligence) MIML bag label set bird song expectation maximization inference image annotation instance annotation instance clustering maximum likelihood multiinstance multilabel learning synthetic datasets Accuracy Birds Clustering algorithms computational modeling Graphical models Logistics Yttrium expectation-maximisation algorithm Graphical models Inference mechanisms Clustering algorithms maximum likelihood computational modeling Logistics Rare earths yttrium ANNOTATIONS PURSE Avifauna Bag device mode matching Power grids

来源：评论

学校读者我要写书评

暂无评论

VLSI Architecture Design of 9/7 Discrete Wavelet Transform for image processing 62

VLSI Architecture Design of 9/7 Discrete Wavelet Transform f...

引用

62nd ieee International Midwest symposium on Circuits and Systems (MWSCAS)

作者： Javed, Sadaf Younis, Ch. Jabbar Alam, Mehboob Massoud, Yehia Mirpur Univ Sci & Technol Dept Elect Engn Mirpur 10250 Ajk Pakistan Stevens Inst Technol Sch Syst & Enterprises Hoboken NJ 07030 USA

ISBN: (纸本)9781728127880

In image processing, transform coding de-correlates images to pre-condition them for efficient compression. In this work, we propose VLSI architecture design of a hardware-efficient 9/7 Discrete Wavelet Transform (DWT). The architecture takes advantage of Canonical Signed Digit (CSD) and Distributed Arithmetic (DA) to represent and optimally distribute co-efficients to reduce the number of adder and shift registers. In addition, the co-efficient multiplication also exploits the horizontal and vertical redundancy in the architecture to reduce the hardware computational complexity. The result is a filter-based design, exploiting hardware path of architecture using CSD coefficients, which finds minimum realization. The proposed architecture is simulated using Verilog Hardware Description Language (HDL). A comparison with other architectures of 9/7 DWT shows a 18.75% reduction in hardware. The result is a hardware-efficient architecture, which provides a low-power solution for image and signal processing applications.

关键词： Discrete Wavelet Transform VLSI Distributed Arithmetic image processing Canonical Signed Digit

来源：评论

学校读者我要写书评

暂无评论

Guest Editorial: AI and Machine Learning Solution Cyber intelligence Technologies: New Methodologies and Applications

引用

ieee TRANSACTIONS ON INDUSTRIAL INFORMATICS 2020年第10期16卷 6626-6631页

作者： Yan, Ke Liu, Lu Xiang, Yong Jin, Qun China Jiliang Univ Coll Informat Engn Hangzhou 310018 Peoples R China Univ Leicester Sch Informat Leicester LE1 7RH Leics England Deakin Univ Sch Informat Technol Burwood Vic 3125 Australia Waseda Univ Dept Human Informat & Cognit Sci Tokorozawa Saitama 3591192 Japan

The papers in this special section focus on new methodologies and applications in artificial intelligence (AI) and machine learning (ML). With the recent development of machine learning (ML), artificial intelligence (AI) and cyber technologies in the field of industrial informatics, it is important to migrate the traditional businesses and services in the physical world to the digital cyber-enabled world. Cyber intelligence technologies, such as the fifth-generation (5G) mobile communication network, big data, Internet of Things (IoT), cloud computing, cognitive computing, ubiquitous computing and blockchains, enable goods, houses, services, information, and capitalization to be shared through the Internet of the Appendix]. Industrial applications, including various mechanical systems, utilities, supply chains, energy systems, power grids, infrastructures, manufactures, traffics, healthcare, and environmental issues, are partially operated or managed remotely with the influence of AI and cyber intelligence technology.

关键词： Special issues and sections Informatics Cloud computing 5G mobile communication Unmanned aerial vehicles Machine learning Artificial intelligence Deep learning

来源：评论

学校读者我要写书评

暂无评论

AUTOMATIC RADIOGRAPHIC QUANTIFICATION OF JOINT SPACE NARROWING PROGRESSION IN RHEUMATOID ARTHRITIS USING POC 16

AUTOMATIC RADIOGRAPHIC QUANTIFICATION OF JOINT SPACE NARROWI...

引用

16th ieee International symposium on Biomedical Imaging (ISBI)

作者： Ou, Yafei Ambalathankandy, Prasoon Shimada, Takeshi Kamishima, Thmotsu Ikebe, Masayuki Hokkaido Univ Res Ctr Integrated Quantum Elect Sapporo Hokkaido Japan Hokkaido Univ Grad Sch Hlth Sci Sapporo Hokkaido Japan

ISBN: (纸本)9781538636411

This paper is an application of image processing techniques for computer-aided diagnosis of Rheumatoid Arthritis (RA). Accurately measuring the progression of joint space narrowing (JSN) is crucial during medical treatment and in imaging biomarkers in clinical trials. In this paper, we analyze sequential radiographic images of patients who have rheumatoid arthritis in hands using image processing techniques. Phase only correlation (POC) is used to detect the progression of JSN between images. A new image processing algorithm is proposed to segment joint images so as to eliminate the mutual interference when measuring the movement of the upper and lower hones by POC. We found that the texture feature on bones will greatly affect the accuracy of POC. Median filter is used to eliminate the effect of texture, and excellent results are obtained in practice. Additionally, the progress of JSN is measured accurately in our method. This can be beneficial for doctors in the identification of disease stages.

关键词： Rheumatoid arthritis Progression of joint space narrowing Phase only correlation Computer-aided diagnosis Automatic quantification image segmentation

来源：评论

学校读者我要写书评

暂无评论

RADAR WAVEFORM RECOGNITION USING FOURIER-BASED SYNCHROSQUEEZING TRANSFORM AND CNN 8

RADAR WAVEFORM RECOGNITION USING FOURIER-BASED SYNCHROSQUEEZ...

引用

8th ieee International Workshop on computational Advances in Multi-Sensor Adaptive processing (CAMSAP)

作者： Kong, Gyuyeol Koivunen, Visa Aalto Univ Dept Signal Proc & Acoust Espoo Finland

ISBN: (纸本)9781728155494

In this paper the problem of recognizing radar waveforms is addressed. Waveform classification is needed in spectrum sharing and radar-communications coexistence, cognitive radars and signal intelligence. Different radar waveforms exhibit different properties in time-frequency domain. We propose a deep learning method for waveform classification. The received signal is processed with Fourier synchrosqueezing transform that has excellent properties in revealing time-varying behavior, rate of, strength and number of oscillatory components in received signals. The resulting time-frequency description is represented as a bivariate image that is fed into a convolutional neural network. The proposed method has superior performance over the widely used the Choi-Williams distribution (CWD) method in distinguishing the polyphase waveforms even at low signal-to-noise ratio regime.

关键词： Choi-Williams distribution convolutional neural network Fourier-based synchrosqueezing transform radar waveform recognition

来源：评论

学校读者我要写书评

暂无评论

Video Face Replacement System Using a Modified Poisson Blending Technique

Video Face Replacement System Using a Modified Poisson Blend...

引用

International symposium on Intelligent signal processing and Communication Systems (ISPACS)

作者： Afifi, Mahmoud Hussain, Khaled F. Ibrahim, Hosny M. Omar, Nagwa M. Assiut Univ Dept Informat Technol Assiut Egypt Assiut Univ Dept Comp Sci Assiut Egypt

ISBN: (纸本)9781479961207

In this paper, we present a system for video face replacement that requires only two videos of a source actor and a target actor using only a single digital camera. Existing video face replacement techniques usually need special equipment or 3D models;the proposed system achieves a realistic replacement of faces without using 3D models or special equipment. There are many applications of the proposed system that are presented in this paper using only two footages of actors. We can replace a frontal face with another one;this gives the possibility to change the appearance of actors without makeup or any prior settings. We introduce a new technique for face blending that is based on a gradient domain method called Modified Poisson Blending ( MPB) to reduce the bleeding problem of Poisson image editing, and achieve realistic results of face replacement. Experimental results demonstrate that the proposed system using MPB technique produces more realistic results than the results of other cloning techniques.

关键词： image processing image cloning interactive image editing face replacement video compositing

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：