東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Visual data detection, modeling and ...

Zhu, Ying.

Linked to FindBook

Google Book

Amazon

博客來

Visual data detection, modeling and enhancement.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Visual data detection, modeling and enhancement./
Author:	Zhu, Ying.
Description:	194 p.
Notes:	Adviser: Stuart Schwartz.
Contained By:	Dissertation Abstracts International64-02B.
Subject:	Computer Science. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3080045

Visual data detection, modeling and enhancement.
Zhu, Ying.

Visual data detection, modeling and enhancement. - 194 p.

Adviser: Stuart Schwartz.

Thesis (Ph.D.)--Princeton University, 2003.

This thesis presents new methods and algorithms for fast face detection, statistical data modeling and image enhancement. Face detection is an important basic technique for a variety of image processing tasks. To improve its applicability, we study fast detection techniques. An efficient framework of multiscale sequential detection is developed to achieve accurate detection with very low complexity. Three useful techniques are employed to speed up the detection: wavelet image modeling with histograms, discriminative feature selection and sequential Bayesian detection. In particular, we introduce several discriminant analysis methods to separate complex class distributions. The technique of error analysis is proposed to select discriminative features adaptively. With adaptive feature selection, a detection based tracking algorithm is also developed. In the second part of the thesis, we present a new paradigm for appearance modeling. A nonlinear generative model is introduced to characterize manifold distributions of deformable patterns. In this model, the visual data is represented with an augmented set of deformable local components. The random deformation is described by probability models. The techniques of basis selection and progressive density estimation are proposed to obtain a data driven representation as well as a manifold distribution model. We use the nonlinear model to characterize handwritten digits and facial expression, construct pose manifolds and derive a layered representation for video data. The last part of the thesis addresses two problems of image quality enhancement: wavelet domain image interpolation and error concealment. A linear composite MMSE estimator is proposed for image interpolation, where a parametric edge model is used to infer the detailed wavelet coefficients, and local statistics are used to minimize the estimation error. For error concealment, we introduce a directional smoothness measure to evaluate the structural consistency around image edges. A block-based concealment algorithm is developed to reconstruct damaged areas and assure consistent edge structures.Subjects--Topical Terms:

626642
Computer Science.

Visual data detection, modeling and enhancement.
LDR:02957nam 2200265 a 45 001 937252
005 20110511
008 110511s2003 eng d
035 $a (UnM)AAI3080045
035 $a AAI3080045
040 $a UnM $c UnM
100 1 $a Zhu, Ying. $3 1261116
245 1 0 $a Visual data detection, modeling and enhancement.
300 $a 194 p.
500 $a Adviser: Stuart Schwartz.
500 $a Source: Dissertation Abstracts International, Volume: 64-02, Section: B, page: 0893.
502 $a Thesis (Ph.D.)--Princeton University, 2003.
520 $a This thesis presents new methods and algorithms for fast face detection, statistical data modeling and image enhancement. Face detection is an important basic technique for a variety of image processing tasks. To improve its applicability, we study fast detection techniques. An efficient framework of multiscale sequential detection is developed to achieve accurate detection with very low complexity. Three useful techniques are employed to speed up the detection: wavelet image modeling with histograms, discriminative feature selection and sequential Bayesian detection. In particular, we introduce several discriminant analysis methods to separate complex class distributions. The technique of error analysis is proposed to select discriminative features adaptively. With adaptive feature selection, a detection based tracking algorithm is also developed. In the second part of the thesis, we present a new paradigm for appearance modeling. A nonlinear generative model is introduced to characterize manifold distributions of deformable patterns. In this model, the visual data is represented with an augmented set of deformable local components. The random deformation is described by probability models. The techniques of basis selection and progressive density estimation are proposed to obtain a data driven representation as well as a manifold distribution model. We use the nonlinear model to characterize handwritten digits and facial expression, construct pose manifolds and derive a layered representation for video data. The last part of the thesis addresses two problems of image quality enhancement: wavelet domain image interpolation and error concealment. A linear composite MMSE estimator is proposed for image interpolation, where a parametric edge model is used to infer the detailed wavelet coefficients, and local statistics are used to minimize the estimation error. For error concealment, we introduce a directional smoothness measure to evaluate the structural consistency around image edges. A block-based concealment algorithm is developed to reconstruct damaged areas and assure consistent edge structures.
590 $a School code: 0181.
650 4 $a Computer Science. $3 626642
650 4 $a Engineering, Electronics and Electrical. $3 626636
690 $a 0544
690 $a 0984
710 2 0 $a Princeton University. $3 645579
773 0 $t Dissertation Abstracts International $g 64-02B.
790 $a 0181
790 1 0 $a Schwartz, Stuart, $e advisor
791 $a Ph.D.
792 $a 2003
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3080045