检索结果-内蒙古大学图书馆

Image quality assessment metrics by using directional projection

Chinese Optics Letters 2008年第7期6卷 491-494页

作者：庞建新张荣张晖黄轩刘政凯 MOE-Microsoft Key Laboratory of Multimedia Computing and Communication Department of Electronic Engineering and Information ScienceUniversity of Science and Technology of ChinaHefei 230027

Objective image quality measure, which is a fundamental and challenging job in image processing, evaluates the image quality consistently with human perception automatically. On the assumption that any image distortion could be modeled as the difference between the directional projection-based maps of reference and distortion images, we propose a new objective quality assessment method based on directional projection for full reference model. Experimental results show that the proposed metrics are well consistent with the subjective quality score.

关键词： PSNR SVD JPEG Image quality assessment metrics by using directional projection DMOS

来源：评论

学校读者我要写书评

暂无评论

On-site travel path recommendation based on fast matching

引用

Journal of Convergence Information Technology 2012年第10期7卷 85-94页

作者： Yin, Huagang Jiang, Kai Yu, Nenghai MOE-Microsoft Key Lab Of Multimedia Computing and Communication University of Science and Technology of China China

Trip planning is generally a very time-consuming task due to the complex trip requirement and the lack of convenient tools/systems to assist the planning. In this paper, we propose an on-site travel path recommendation service based on geo-tagged photos to facilitate tourists' trip planning. The large scale geo-tagged photos that are publically available on the web make this service possible, as geotagged photos encode rich travel-related metadata and can be used to mine travel paths from previous tourists. In this work, about 20 million geo-tagged photos were crawled from ***. Then a substantial number of travel paths are minded from the crawled geo-tagged photos. After that, a search system is built to index and search paths, and Directed Sparse Chamfer Distance is proposed to measure the similarity of two paths in linear time. Based on 20 million geo-tagged photos, the proposed service is evaluated using both objective and subjective evaluation methods and shows promising results.

关键词： Trip Planning Geo-Tagged Photos Path Mining

来源：评论

学校读者我要写书评

暂无评论

Directly modeling of correlation matrices for GMM in speaker identification

Directly modeling of correlation matrices for GMM in speaker...

引用

18th International Conference on Pattern Recognition (ICPR 2006)

作者： Yao, Zhiqiang Zhou, Xi Dai, Beiqian Liu, Minghui Xie, Yanlu MOE Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China China

ISBN: (纸本)0769525210

In this paper, we present a new framework to model full covariance matrices of Gaussian components. In this framework, directly modeling the full correlation matrix instead of the full covariance matrix is our purpose, as the correlation matrix is the direct description of the correlation of inter feature elements. In order to model full correlation matrices, we share linear transformations among components' full correlation matrices. Thus, the full correlation matrix of each component is represented by a shared linear transformation and a component-specific diagonal correlation matrix. The transformation is used to help the diagonal correlation matrix to model the correlation of inter feature-vector elements more precisely. We evaluate our new framework on a mandarin speaker identification task. Experiments show that above 35% reduction in speaker identification error rate is achieved compared with the best diagonal covariance models. Furthermore, our algorithm achieved better performance than STC does.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Improved two-stage Wiener filter for robust speaker identification

Improved two-stage Wiener filter for robust speaker identifi...

引用

18th International Conference on Pattern Recognition, ICPR 2006

作者： Yanlu, Xie Minghui, Liu Zhiqiang, Yao Beiqian, Dai MOE Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China

In order to solve the problem of robustness in textindependent speaker identification, two-stage Wiener filter presented in the ETSI standard is used in this paper. However, the performance of the voice activity detection (VAD) block in the standard is not so satisfying under lower SNR. Thus constrained-clustering vector quantization is presented to improve the block in the Wiener filter method. Experiments based on the MSRA 's mandarin speech corpora show the good robustness of the improved method to several common additive noises. © 2006 IEEE.

关键词： Speech recognition

来源：评论

学校读者我要写书评

暂无评论

Modeling glottal source for high quality voice conversion

Modeling glottal source for high quality voice conversion

引用

6th World Congress on Intelligent Control and Automation, WCICA 2006

作者： Jun, Sun Beiqian, Dai Jian, Zhang Yanlu, Xie MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China

ISBN: (纸本)1424403324

A novel modeling method for glottal source is proposed for Improving the naturalness and quality of synthetic speech. This paper utilizes the high correlation between vocal tract parameters and glottal source to model glottal source. Vocal tract parameters (LSF) are clustered into some classes. Within each class, a LSF vector closest to centroid and Its corresponding glottal wave derivative are selected as a code vector representing different phonetic class of voiced speech. At the stage of voice conversion or synthesis, we can find the relevant glottal source by virtual of finding the closest matched vocal tract parameters. Experiment results show that this vocal tract related glottal source model significantly outperform Rosenberg model and LF model. Correlation coefficients between vocal tract related glottal source and original glottal source increase 27% and 30.13%, spectral distance between synthetic speech and original speech reduce 50.5% and 51.48% respectively, comparing with Rosenberg model and LF model. © 2006 IEEE.

关键词： Speech synthesis

来源：评论

学校读者我要写书评

暂无评论

Kurtosis normalization in feature space for robust speaker verification

Kurtosis normalization in feature space for robust speaker v...

引用

2006 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2006

作者： Yanlu, Xie Beiqian, Dai Zhiqiang, Yao Minghui, Liu MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China

ISBN: (纸本)142440469X

The acoustic mismatch between the training and test environments will lead to the difference of the statistical characteristics of speech parameters. Since the statistical characteristics of the kurtosis can measure the non-Gaussianity of a random variable, kurtosis normalization will make the training and test speech parameters match the standard normal distribution in some sense. In this paper, a kurtosis normalization method using sigmoid functions (logit functions) in feature space is presented for GMM-UBM based text-independent speaker verification system. Experimental results on the 2004 NIST SRE database show that with the new method significant improvement can be achieved in not only equal error rate but also minimum detection cost compared with baseline system (more than 33% relative reduction for long speech). © 2006 IEEE.

关键词： Feature extraction

来源：评论

学校读者我要写书评

暂无评论

PROMPTTTS 2: DESCRIBING AND GENERATING VOICES WITH TEXT PROMPT 12

PROMPTTTS 2: DESCRIBING AND GENERATING VOICES WITH TEXT PROM...

引用

12th International Conference on Learning Representations, ICLR 2024

作者： Leng, Yichong Guo, Zhifang Shen, Kai Ju, Zeqian Tan, Xu Liu, Yanqing Liu, Yufei Yang, Dongchao Zhang, Leying Song, Kaitao He, Lei Li, Xiang-Yang Zhao, Sheng Qin, Tao Bian, Jiang MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Microsoft China

Speech conveys more information than text, as the same word can be uttered in various voices to convey diverse information. Compared to traditional text-to-speech (TTS) methods relying on speech prompts (reference speech) for voice variability, using text prompts (descriptions) is more user-friendly since speech prompts can be hard to find or may not exist at all. TTS approaches based on the text prompt face two main challenges: 1) the one-to-many problem, where not all details about voice variability can be described in the text prompt, and 2) the limited availability of text prompt datasets, where vendors and large cost of data labeling are required to write text prompts for speech. In this work, we introduce PromptTTS 2 to address these challenges with a variation network to provide variability information of voice not captured by text prompts, and a prompt generation pipeline to utilize the large language models (LLM) to compose high quality text prompts. Specifically, the variation network predicts the representation extracted from the reference speech (which contains full information about voice variability) based on the text prompt representation. For the prompt generation pipeline, it generates text prompts for speech with a speech language understanding model to recognize voice attributes (e.g., gender, speed) from speech and a large language model to formulate text prompts based on the recognition results. Experiments on a large-scale (44K hours) speech dataset demonstrate that compared to the previous works, PromptTTS 2 generates voices more consistent with text prompts and supports the sampling of diverse voice variability, thereby offering users more choices on voice generation. Additionally, the prompt generation pipeline produces high-quality text prompts, eliminating the large labeling cost. The demo page of PromptTTS 2 is available. © 2024 12th International Conference on Learning Representations, ICLR 2024. All rights reserved.

关键词： Large datasets

来源：评论

学校读者我要写书评

暂无评论

A ROBUST EYE LOCALIZATION ALGORITHM FOR FACE RECOGNITION

引用

Journal of Electronics(China) 2008年第3期25卷 337-342页

作者： Zhang Wencong Li Xin Yao Peng Li Bin Zhuang Zhenquan MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Hefei 230026 China

The accuracy of face alignment affects greatly the performance of a face recognition system. Since the face alignment is usually conducted using eye positions, the algorithm for accurate eye lo- calization is essential for the accurate face recognition. In this paper, an algorithm is proposed for eye localization. First, the proper AdaBoost detection is adaptively trained to segment the region based on the special gray distribution in the region. After that, a fast radial symmetry operator is used to pre- cisely locate the center of eyes. Experimental results show that the method can accurately locate the eyes, and it is robust to the variations of face poses, illuminations, expressions, and accessories.

关键词： Eye localization AdaBoost Fast radial symmetry

来源：评论

学校读者我要写书评

暂无评论

Tracking based on SURF and superpixel

Tracking based on SURF and superpixel

引用

International Conference on Image and Graphics

作者： Liu, Yu Zhou, Wei Yin, Huagang Yu, Nenghai MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Hefei China

ISBN: (纸本)9780769545417

In this paper we present a novel algorithm for object tracking in video sequence based on SURF key-point and superpixel. SURF key-point is very effective for object matching between two images and we can use it to locate object by and large. But only in this way it can't guarantee the success of object tracking because many available key-points are un-matched and can't be used to locate the object. In order to solve this problem we take the advantage of superpixel to obtain more available key-points based on its property that all points in the same superpixel region belong to the same object. At last we construct a weight map for each frame to refine the location of target *** experiments demonstrate that our work is robust for tracking, especially dealing with occlusions and object transformations. © 2011 IEEE.

关键词： Superpixels

来源：评论

学校读者我要写书评

暂无评论

Online travel destination recommendation with efficient variable memory Markov model

Online travel destination recommendation with efficient vari...

引用

2013 IEEE International Conference on multimedia and Expo Workshops, ICMEW 2013

作者： Kai, Jiang Yu, Nenghai Li, Weihai MOE-Microsoft Key Laboratory of Multimedia Computing and Communication University of Science and Technology of China Hefei China

ISBN: (纸本)9781479916047

Online travel destination recommendation is to keep track of a user's current traveling history to recommend next destination in real time while the user is on the travel. This paper presents an efficient variable memory Markov model based method to provide such recommendation. The proposed method utilizes the large quantity of geotags from photo sharing website and combines travel pattern, location's popularity and distance factors to generate real time recommendation. Experiments on Panoramio data set demonstrate the effectiveness of this method. © 2013 IEEE.

关键词： Social networking (online)

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：