Wavelet multiresolution and applications.
Item
-
Title
-
Wavelet multiresolution and applications.
-
Identifier
-
AAI9618124
-
identifier
-
9618124
-
Creator
-
Zhu, Jianxin.
-
Contributor
-
Adviser: Joseph Barba
-
Date
-
1996
-
Language
-
English
-
Publisher
-
City University of New York.
-
Subject
-
Engineering, Biomedical
-
Abstract
-
Multiresolution analysis represents a function f(x) at various levels ot resolution. It is more efficient to analyze the image at resolution {dollar}R\sb{lcub}j{rcub},{dollar} and then process the additional details available at the resolution {dollar}R\sb{lcub}j+1{rcub}.{dollar} Wavelet representation provides a coarse approximation of the signal plus the detail signals at the successive resolutions {dollar}R\sb{lcub}j{rcub}{dollar} for {dollar}1 \le j \le N.{dollar} In this dissertation, wavelet functions, wavelet bases and methods to construct wavelet bases are presented. The major contribution of this research is in the application of wavelet for voice pitch detection, cell contour extraction, and texture segmentation. In wavelet multiresolution voice application, detail signals on different scales are used to detect the voice signal pitch period. In cell image application, difference signals are used to extract cell contours. Edge enhancement method provide another way to extract cell contours. Different contour extraction algorithms are developed and the experimental results are demonstrated. Wavelet energy representation based on multichannel decomposition is defined. A pyramid tree structure and a quadtree structure are used to construct energy vectors which are used as the texture feature in segmentation. Test images, with structured/unstructured and rotated textures are applied to our segmentation algorithm. Experimental results are shown to demonstrate that the algorithms are very successful.
-
Type
-
dissertation
-
Source
-
PQT Legacy CUNY.xlsx
-
degree
-
Ph.D.
