A Survey On Astronomical Image Deconvolution

PhD Qualifying Examination


Title: "A Survey On Astronomical Image Deconvolution"

by

Miss Yixin LIU


Abstract:

Image deconvolution is a technique to remove noise and sharpen the 
contrast in a “dirty” image, and has been applied on widefield as well as 
microscopic images to improve image quality. This survey focuses on radio 
astronomical images, which are generated from radio telescopes. 
Representative deconvolution algorithms on these images are either 
iterative or statistical. Iterative methods select a subset of pixels to 
clean in each iteration whereas statistical algorithms build a 
mathematical model to fit the entire image. As a result, iterative methods 
work well on images of point sources and statistical algorithms excel on 
those of extended objects. However, statistical algorithms are 
computationally more expensive and consume more memory than iterative 
methods. As both point sources and extended objects are studied in radio 
astronomical images, we discuss possible improvements on both kinds of 
deconvolution methods, for example, adding new features in iterative 
algorithms to enhance quality, and parallelizing statistical methods to 
improve time efficiency.


Date:  			Monday, 13 June 2022

Time:                  	2:00pm - 4:00pm

Zoom Meeting: 
https://hkust.zoom.us/j/98126430868?pwd=UitzbDI1akc5eEN6Y3BNNE5CVTJIdz09

Committee Members:	Prof. Qiong Luo (Supervisor)
 			Prof. Pedro Sander (Chairperson)
 			Dr. Qifeng Chen
 			Dr. Dan Xu


**** ALL are Welcome ****