Wavelet polynomial threshold based filter for high resolution microscopy

Date
2011
Authors
Chan, Michael
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract

Bioimaging is becoming an extensive application area for telemedicine, other emerging medical fields and non-diagnostic research explorations. Bioimaging refers to various types and scales imagery capture using diverse types of instrumentation such as microscopes, computed tomography (CT), magnetic resonance imaging (MRI), etc. For example, neuroimaging has been used to understand human brain and even apply similar concepts for human-machine interface development.

Resolution enhancement is used for manipulating or exploring medical images. Significant number of low resolution images has been generated in the past due to old-style equipment, compression for storage or transmission, or because high resolution instruments are typically more costly and less widely deployed. Magnification for better observation helps physicians to make accurate diagnosis. Conventional zooming typically enlarges images visually, but it does not improve the displaying/resolution quality, it does not enhance details and reduce visual degradations.

In this thesis the following problem is investigated: to what extent the quality of low resolution imagery can be enhanced by filtering compressed data or applying the interpolation and filtering chain. A novel optimal filtering concept, with and without interpolation, is used for increasing the resolution of microscopic images and compression artifact removal. The fundamental principle is based on a polynomial thresholding concept in transform domain, particularly when using wavelet transforms. Optimization of such a threshold provides a balancing trade-off between various conventional techniques such as soft and hard thresholding. This technique helps to address systematically biased estimates in soft thresholding, and poorer denoising performance of hard thresholding estimates. Various related ideas are investigated. The proposed approach demonstrates promising results for improving resolution of biomedical images which is validated through simulations.

Description
This item is available only to currently enrolled UTSA students, faculty or staff.
Keywords
Bioimaging, Denoising, Microscopy, Resolution, Signal Processing, Wavelet
Citation
Department
Electrical and Computer Engineering